3D Laser Scanning for HeritageAdvice and guidance to users on laser scanning in archaeology and architecture

2007

2

Contents

1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31.1 Aims . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31.2 The Heritage3D project . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31.3 Three-dimensional recording . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31.4 Questions laser scanning can help to answer . . . . . . . . . . . . . . . . . . . . . . . . . . 41.5 Tasks appropriate for laser scanning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41.6 What laser scanning cannot provide . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

2 How does laser scanning work? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72.1 Instrumentation and hardware . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72.2 Software . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92.3 Computer hardware . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

3 Commissioning survey . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93.1 Knowing what you want . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93.2 Determining appropriate point density . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103.3 Finding a contractor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103.4 Laser safety . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103.5 Archived data sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

4 From point cloud to useful information . . . . . . . . . . . . . . . . . . . . . . . . . 114.1 Typical workflows . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114.2 Cloud alignment/registration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124.3 Modelling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124.4 Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

5 Managing data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155.1 Reprocessing data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155.2 Data formats and archiving . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155.3 Metadata . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16

6 Helping you to decide . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166.1 What outputs are wanted? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166.2 How big is the subject? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166.3 What level of accuracy is required? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166.4 What resolution of measurement? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166.5 Does the survey need to be referenced? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166.6 Time and access restrictions? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166.7 Is three-dimensional information required? . . . . . . . . . . . . . . . . . . . . . . . . . . 176.8 Budget . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176.9 Can you do this yourself? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176.10 What are the alternatives? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17

7 Where to find out more . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187.1 Charters and guidance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187.2 Organisations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187.3 Books . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187.4 Journals and conference proceedings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187.5 Websites . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187.6 Training . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

8 Glossary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

9 Case study summaries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20Case study contributors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42

Project overview

The Heritage3D project directly addresses four sections of the 1998English Heritage Exploring our PastImplementation plan. The two principalaims of the project are to:

Develop and support best practice inlaser scanning for archaeology andarchitecture

Disseminate this best practice to usersalong with the education of likelybeneficiaries

In order to achieve these aims the projectworks towards five objectives:

Objective 1 – production of a guidancenote that demonstrates the products thatcan be generated from laser scanning

Objective 2 – to update the currentAddendum to the Metric SurveySpecification to take into account thecontinuing advances in the technology

Objective 3 – to increase the knowledgebase of English Heritage by formingpartnerships with external surveypractitioners/equipment manufacturerswithin the UK

Objective 4 – to promote synthesisbetween disciplines within EnglishHeritage by publishing and maintaining aproject website

Objective 5 – to provide workshops onthe use of laser scanning to educatearchaeologists, architects and engineersfrom within English Heritage

1 Introduction

1.1 AimsThe advice and guidance presented hereaims to provide the reader with theinformation they require to use laserscanning appropriately and successfullywithin projects. However, it should benoted that other survey techniques canprovide three-dimensional informationand should be considered alongside laserscanning. So while this document presents information specifically on when,why and how you might want to use laserscanning, it will point to other techniquesthat might also be considered. Moreover,it will cover generic issues, such as datamanagement, where the advice andguidance given will be relevant to anygeometric survey techniques. As a result of this note users should be able tounderstand how laser scanning works,why they might need to use it and how itmight be applied.

For abbreviations see Glossary.

1.2 The Heritage3D projectThis note has been generated as part ofthe Heritage3D project. Heritage3D wassponsored by English Heritage’s HistoricEnvironment Enabling Programme(project 3789 MAIN) and undertaken by the School of Civil Engineering andGeosciences at Newcastle University.This two year project developed andsupported best practice in laser scanningfor archaeology and architecture, anddisseminated this best practice to users.Further details on the project can befound at the Heritage3D websitehttp://www.heritage3d.org. A summaryof the case studies referred to throughoutthis note is given at the end of thisdocument.

1.3 Three-dimensional recordingThe recording of position, dimensionsand/or shape is a necessary part of almostevery project related to the conservationof cultural heritage, forming an importantelement of the documentation andanalysis process. For example, knowingthe size and shape of a topographicfeature located in a historic landscape canhelp archaeologists identify itssignificance, knowing how quickly a stonecarving is eroding helps a conservator todetermine the appropriate action for itsprotection, while simply having access to aclear and accurate record of a buildingfaçade helps a project manager toschedule the work for its restoration.

It is common to present suchmeasurements as plans, sections and/orprofiles plotted onto hardcopy for directuse on site. However, with theintroduction of new methods for three-dimensional measurement and increasingcomputer literacy among users, there is agrowing demand for three-dimensionaldigital information.

There is a wide variety of techniques forthree-dimensional measurement. Thesetechniques can be characterised by thescale at which they might be used (whichis related to the size of the object theycould be used to measure), and on thenumber of measurements they might beused to acquire (which is related to thecomplexity of the object). Figure 1summarises these techniques in terms ofscale and object complexity. While handmeasurements can provide dimensions andposition over a few metres, it is impracticalto extend this to larger objects; andcollecting many measurements (forexample 1000 or more) would be alaborious and, therefore, unattractiveprocess. Photogrammetry and laserscanning could be used to provide agreater number of measurements forsimilar object sizes, and, therefore, aresuitable for more complex objects.Photogrammetry and laser scanning mayalso be deployed from the air so as toprovide survey data covering much largerareas. While GPS might be used to surveysimilarly sized areas, the number of pointsit might be used to collect is limited when

compared to airborne, or even spaceborne,techniques. This advice and guidance isfocused closely on laser scanning (from theground or air), although the reader shouldalways bear in mind that anothertechnique may be able to provide theinformation required.

Laser scanning, from the air or from theground, is one of those technicaldevelopments that enables a large quantityof three-dimensional measurements to becollected in a short space of time. Thisdocument presents advice and guidanceon the use of laser scanning, so thatarchaeologists, conservators and othercultural heritage professionals can makethe best possible use of this technique.

The term laser scanner applies to a rangeof instruments that operate on differingprinciples, in different environments andwith different levels of accuracy. A genericdefinition of a laser scanner, taken fromBöhler and Marbs is:

“any device that collects 3D co-ordinatesof a given region of an object’s surfaceautomatically and in a systematic patternat a high rate (hundreds or thousands ofpoints per second) achieving the results(ie three-dimensional co-ordinates) in(near) real time.”

(Böhler, W, and Marbs, A 2002, ‘3DScanning Instruments’, Proceedings ofCIPA WG6 Scanning for Cultural HeritageRecording, September 1–2, Corfu, Greece)

3

1 Mil

100 000

10 000

1000

0.1m 1m 10m 100m 1km 10km 100km 1000km

100

10

1

Object size

Com

plex

ity(p

oint

s/ob

ject

)

Close range photogrammetry/laser

scanningAirborne

photogrammetry/laser scanning

Terrestrialsurvey

GPSHand

measurement

Satellite remote sensing

Fig 1 Three-dimensional survey techniques characterised by scale and object size (derived from Böhler presentation CIPAsymposium 2001, Potsdam).

This process might be undertaken from a static position or from a movingplatform, such as an aircraft. Airbornelaser scanning is frequently referred to as LiDAR, although LiDAR is a termthat applies to a particular principle of operation, which includes laserscanners used from the ground. Laserscanning is the preferred generic term and will be used throughout this guide to refer to ground based and airbornesystems.

Laser scanning from any platformgenerates a point cloud: a collection ofXYZ co-ordinates in a common co-ordinate system that portrays to the viewer an understanding of the spatialdistribution of a subject. It may alsoinclude additional information, such aspulse amplitude or RGB values.Generally, a point cloud contains arelatively large number of co-ordinates in comparison with the volume the cloud occupies, rather than a few widelydistributed points.

1.4 Questions laser scanning can help to answerThe key to deciding if you need to uselaser scanning is thinking carefully about the questions you want to answerwithin your project. The sorts of questions that you’ll be asking will varyfrom discipline to discipline. Typicalquestions might be as simple as “Whatdoes it look like?” or “How big is it?”For example, a conservator might want to know how quickly a feature ischanging, while an archaeologist might be interested in understanding how onefeature in the landscape relates to another. An engineer might simply wantto know the size of a structure and where existing services are located.In other terminology, laser scanning might be able to help inform on aparticular subject by contributing to theunderstanding. Scanning may alsoimprove the accessibility of the object.Once you have a clear idea of thequestions you want to answer, thenwhether you need or are able to use laser

scanning will depend on a range ofvariables and constraints.

1.5 Tasks appropriate for laser scanningLaser scanning of all types might have ause at any stage of a project. Tasks that youmight find being considered could include:

l contributing to a record prior torenovation of a subject or site whichwould help in the design process, inaddition to contributing to the archiverecord (see Case Study 11)

l contributing to a detailed record where afeature, structure or site might be lost/changed forever, such as in an archaeo-logical excavation or at a site at risk

l structural or condition monitoring, suchas looking at how the surface of anobject changes over time in response toweather, pollution or vandalism

l providing a digital geometric model from which a replica model may begenerated for display or as a replacementin a restoration scheme (see Case Study 4)

4

Fig 2 (left) Triangulation laser scanning(courtesy of Conservation Technologies,National Museums Liverpool).

Fig 3 (above) Time of flight laser scanning(courtesy of Newcastle University).

Fig 4 (right) Airborne laser scanninginstrumentation (http://www.optech.ca/).

l contributing to three-dimensionalmodels, animations and illustrations forpresentation in visitor centres, museumsand through the media (enhancingaccessibility/engagement and helping toimprove understanding)

l aiding the interpretation of archaeo-logical features and their relationshipacross a landscape, thus contributing tothe understanding about the developmentof a site and its significance to the area

l working, at a variety of scales, to

uncover previously unnoticedarchaeologically significant features suchas tool marks on an artefact, or lookingat a landscape covered in vegetation orwoodland (see Case Study 15)

l spatial analysis, not possible withoutthree-dimensional data, such as line ofsight or exaggeration of elevation

However, it is important to recognise thatlaser scanning is unlikely to be used inisolation to perform these tasks. It is

highly recommended that photographyshould be collected to provide a narrative record of the subject. Inaddition, on-site drawings, existingmapping and other survey measurementsmight also be required. The capture ofadditional data helps to protect a user as it helps to ensure the requiredquestions can be answered as well aspossible, even if the a subject has changed or even been destroyed since its survey (Figs 5–12).

5

Fig 5 (left) An original and replica bust of the Emperor Caligula generated from data collected by a triangulation laser scanner (courtesy of Conservation Technologies, NationalMuseums Liverpool).

Fig 6 (above) Laser scanning for historic sites at risk, St Mary’s Church Whitby, North Yorkshire.

Fig 7 Use of laser scanning data for presentation ofarchaeology: Ketley Crag rock shelter (courtesy of Paul Bryan,English Heritage).

6

Fig 11 Laser scanning contributing to the site record of a Neolithic flint mine in Norfolk(courtesy of Paul Bryan, English Heritage).

Fig 12 Looking at earthworks covered by vegetation (courtesy Simon Crutchley, EnglishHeritage and the Forestry Commission, data provided by Cambridge University Unit forLandscape Modelling).

Fig 8 Profile of point cloud data used for a structural survey (courtesy Tony Rogers, APR Services).

Fig 9 Using laser scanning to contribute to a record during excavation (courtesy of theDiscovery Programme Ltd).

Fig 10 (left) Using airborne laser scanning to understand a historic landscape: a LiDAR imageof the area around Charterhouse Roman town on the Mendip Hills.To the north-west is anamphitheatre (A), to the south-east are faint traces of the Roman road (B). In the bottomcentre is the Roman fortlet (C), not to be confused with the sub-rectangular enclosure (D) of probable medieval or later date overlying the remains of the Roman town.The imageis colour shaded according to height (ranging from red = high to blue = low); the height hasbeen exaggerated to emphasise the features (courtesy of Mendip Hills AONB – Original sourceUnit for Landscape Modelling (ULM) Cambridge University).

1.6 What laser scanning cannot provideLaser scanning will not provide a solutionto all recording tasks. It does not provideunlimited geometric accuracy andcompleteness over objects and landscapesof all sizes at a low cost. In many cases,laser scanning might be consideredunnecessary for the level of ‘deliverable’required. Scanning may also take a longtime to achieve the level of results yourequire.

Laser scanning is not as versatile as acamera, for it requires time to scan theobject, whereas a camera can record ascene in a matter of seconds. Laserscanning cannot see through objects(including dense vegetation), and itcannot see around corners. Scanningsystems have minimum and maximumranges over which they operate. Scanningabove or below these ranges should beavoided so as to prevent inaccurate datacapture. Some laser scanning equipmentcan have problems with certain materialtypes, such as marble or gilded surfaces.

While the point cloud generated by laserscanning may be useful on its own, it ismore than likely that the cloud will be ameans to an end rather than the end itself.Laser scanning is best suited to therecording of surface information, ratherthan edges and discrete points, although it

is increasingly used to generate two-dimensional sections, profiles and planswhere supporting information, such asimagery, is also available.

2 How does laser scanning work?

2.1 Instrumentation and hardwareObviously, particular tasks will havespecific requirements. Generally, the largerthe object the lower the accuracy andresolution that can be achieved

realistically. Laser scanners generallyoperate on one of three principles:triangulation, time of flight or phasecomparison. Table 1 provides a shortsummary of these techniques, includingtypical system accuracy and the typicaloperating ranges. The following discussiondescribes each technique in further detail.

TriangulationTriangulation scanners calculate 3D co-ordinate measurements by triangulatingthe position of a spot or stripe of laser

7

Table 1 Laser scanning techniques used in cultural heritage management activities

scanning system use typical accuracy / operating range

rotation stage scanning small objects (that can be removed from site) 50 microns / 0.1m–1mCan be used to produce data suitable for a replica ofthe object to be made.

triangulation-based artefact arm mounted scanning of small objects and small surfaces 50 microns / 0.1m–1mscanners Can be performed on site it required.

May be used to produce a replica.

mirror/prism scanning small objects surface areas in situ sub-mm / 0.1m–25mCan be used to produce a replica.

terrestrial time of flight Suitable for survey of building façades and interiors 3–6 mm at ranges up to 100m / 2m–100mlaser scanners resulting in line drawings (with supporting data) and

surface models.

terrestrial phase comparison Suitable for survey of building façades and interiors 5mm at ranges up to 2m / 2m–50mlaser scanners resulting in line drawings (with supporting data) and

surface models.

Airborne laser scanning prospecting landforms (including in forested areas) 0.15m (depending on the parameters of the survey) / 10m–3500m

(adapted from Barber, DM, Dallas, RWA and Mills, JP 2006, ‘Laser scanning for architectural conservation’, J Archit Conserv 12, 35–52).

Fig 13 (above) Triangulation laser scanning of rock art, on site with a canopy to reduce the influence of bright sunlight(courtesy of Tertia Barnet).

light. A basic outline of a triangulationsystem is given in Figure 15. Sometriangulation systems require an object tobe placed on a moveable turntable thatrotates the object in front of a staticscanner. Alternatively, triangulationsystems can be mounted on mechanicalarms (Fig 2), which, while site portable,are more often found in specialist studiosor laboratories. Typically, with this type ofsystem, the scanner-to-object distance isless than 0.5m and commonly has ameasurement accuracy of 0.1m. Althoughnot providing the very high level ofaccuracy associated with arm-basedscanners, there are triangulation systemsthat scan the measurement beamautomatically, using mechanised prismsand mirrors. These systems can be likenedto a tripod-based camera used to collectoverlapping three-dimensional images ofthe subject at ranges of up to 2m. Suchsystems tend to be the most portabledesign, and are ideal for recording smallarchitectural features such as detailedcarvings or cut marks. Finally, sometriangulation-based systems enablemeasurements at a range of up to 25m,although at this range you can expect afurther degradation in accuracy.Triangulation scanners typically performbadly in bright sunlight, so temporaryshading is often required (Figs 13–15).

Time of flightSystems based on the measurement of thetime of flight of a laser pulse andappropriate to architectural conservationactivities offer an accuracy of between3mm and 6mm. Such systems use thetwo-way travel time of a pulse of laserenergy to calculate a range.

In comparison to triangulation systems,scanners using the time-of-flight methodare more suited to general architecturalrecording tasks, owing to their longerranges (typically between 2m to 100m).This type of scanner can be expected tocollect many thousands of points everyminute by deflecting this laser pulse acrossan object’s surface, using a rotating mirroror prism (Figs 16–18).

Phase comparisonPhase-comparison systems, while offeringsimilar accuracies to time-of-flightsystems, calculate the range to the targetslightly differently. A phase-comparisonbases its measurement of range on thedifferences in the signal between theemitted and returning laser pulses,rather than on the time of flight. As this is a continuous process, phase-comparison systems have much higherrates of data capture (millions of pointsper minute), which can lead to significantpressures on computer hardware insubsequent processing. Time-of-flight andphase-comparison systems are typicallyable to scan a full 360 degrees in thehorizontal and often up to 180 degrees in the vertical.

Airborne laser scanningAirborne laser scanners use laser scanningequipment based on time-of-flight orphase-comparison principles. However, itis also necessary to couple the laserscanner with sensors to measure theposition, orientation and attitude of theaircraft during data collection, by GPSand inertial sensors. By combining thesemeasurements with the range datacollected by the laser scanner a three-dimensional point cloud representing thetopography of the land is produced, muchlike that generated from a ground-basedstatic scanner (Fig 19).

8

1. A laser generates ameasurement beam

2. A rotating mirror deflectsthis beam across an object

4. A lens system focuses the reflected light on a sensor

5.The location of the laser pulse on the sensor is combinedwith the known seperation between it and the mirror todetermine a point coordinate by triangulation

3.The measurmentbeam is reflected from the object’s surface

Fig 14 A laser stripe from a triangulation scanner.

1. Laser/receiving optics generates a pulse and starts a timer

2. Rotating mirror deflects the measurment beam

4. Reflected pulse is returningto the receiving optics and the timer is stopped. This time offlight, the known speen of light,and mirror angles are used to determine the XYZ coordinates

Distance

3. Laser strikesthe objectand is reflected

Fig 16 A time-of-flight laser scanner, showing measurementbeam and direction of scan (courtesy of Riegl UK).

Fig 18 Measurement using a time-of-flight or a phase-comparison measurement system.

Fig 17 A phase-comparison scanner (courtesy of Z+F UK).

Fig 15 A schematic of a mirror based triangulationmeasurement system.

2.2 SoftwareComputer software is required at eachstage of the laser scanning process. Thisincludes the operation of the scanner, theprocessing of the collected data and thevisualisation and utilisation of thedelivered digital product. Operation of thescanner is likely to be handled by acontractor. In this discussion we willrestrict ourselves to describing softwarefor processing the collected data (alsolikely to be done by the contractor, butgiven here to provide an overview), andsoftware that a user may need for usingthe final results.

The choice of software will be based on anumber of factors, including data quantity,the type of ‘deliverable’ required and userexpertise and skill. The process of turninga point cloud into useful information iscovered in Section 5 below. However, it isuseful here to highlight the significantcomponents of software specially designedto be used with point cloud data.

Such software will offer a three-dimensional viewer that can be used topreview the dataset. It will allow the viewto be rotated, zoomed and panned,colours to be changed and data to beclipped from view. The software will havebeen designed specifically to handle largevolumes of three-dimensionalmeasurements. Mainstream software forCAD, GIS or 3D modelling may not bedesigned to handling the large datasetsgenerated by laser scanning, although insome cases specialist tools can be obtainedto improve the performance of thesemainstream tools, allowing the use of afamiliar software environment.

A user who is commissioning a laserscanning survey is unlikely to need toconsider exactly what software to use toprocess the collected data; rather, he/shewill need to ensure that the methodology isappropriate for their needs. The user will,however, need to ensure that the finalproduct, generated from the point cloud,can be used for the task intended. He/shemay want to manipulate this with astandard desktop GIS package, or mayrequire specialist software to enable easiervisualisation and analysis. Free viewersdesigned for standard and proprietaryformats are available, and low cost tools,designed to give a little more flexibility(such as the ability to make simplemeasurements) can be purchased. For moreinformation on particular products, seebelow, section 7 Where to find out more.

2.3 Computer hardwareA standard desktop PC designed forstandard office use may be insufficient totake full advantage of the generatedproduct, or for the analysis you wish tocarry out. However, desktop PCs withcomputing power and specificationssuitable for the day-to-day use of largegeometric models (assuming appropriatesoftware is available) are more accessibleand less expensive now. At the time ofwriting, if you are planning to buy a newmachine or upgrade an existing one inpreparation for the use of three-dimensional data, consider the following:

l 3D graphics acceleration: Having adedicated 3D graphics card is one of themost important features. Choose onewith at least 256 MB of dedicated videoRAM. Note, some off-the-shelfmachines provide 3D accelerationthrough integrated cards that share thecomputer’s standard memory. Althoughless expensive this type of card shouldbe avoided.

l RAM: Plan to have at least 1 GB ofRAM, although the more the better.Memory is normally installed in pairs ofmodules, so if you are buying a newmachine, consider what will be the mostcost-effective way to add more memoryin the future.

l Hard disk: At least 100 GB will berequired for day-to-day storage.Consider using an external USB harddisk to provide a local backup. At thetime of writing, external USB disks witha 300 GB capacity cost as little as £100.

l Display: Do not underestimate the valueof choosing a good quality monitor. Ifyou have desk space, consider using a

CRT version rather than the morepopular flat-screen LCDs, as CRTscreens often give a much better image,and are less expensive than theirequivalent TFT versions.

l Processor speed and type: While havinga fast processor may improve generalperformance, it is less important thanare graphics card and RAM. Usersshould aim for a processor speed of atleast 2 GHz.

While it may seem expensive to buy awhole new system, an existing desktop PCmight be upgradeable by the simpleaddition of some extra RAM, a newgraphics card and an additional hard drive(changes that might cost less than £300 atthe time of writing).

Do not forget that whatever software youchoose to manipulate the derived models,you may also benefit from some training.Dedicated training helps to get youstarted on the right foot and stops youfrom adopting bad practices early on.Software developers, service providers orsuitable educational establishments mayall be able to provide appropriate training;for organisations that may be able tosuggest suitable training partners, seebelow, section 7 Where to find out more.

3 Commissioning survey

3.1 Knowing what you wantIt is unlikely that an individual requiringlaser scanning will have the means orexpertise to undertake the work him- orherself. It is more likely that survey workwill need to be commissioned andundertaken by a specialist contractor. Thefollowing tips will help you whenpreparing to commission a survey:

l Consider the level of detail required andthe extent of the object/area. These areoften the overriding parameters used todetermine appropriate survey techniqueand/or deliverable product.

l Start by working out what you need inorder to answer the questions you haveset. Try to come up with a requirementfor accuracy and product. Realise thatyou might not need to specify the actualtechnique, just the product you require.

l Consider how you will use the productbefore it is procured/commissioned;additional costs might be hidden inbuying new software/hardware.

l Discuss the requirements with possible contractors. A good contractor

9

Fig 19 Airborne laser scanning.

will be able to advise you if yourrequirements are achievable. Alsodiscuss the work with other members of the team, especially with those whoseexpertise is greater than yours, as other uses of the survey may be moreapparent to them, and may increase the overall value of the work to becommissioned.

l Consider how the collected survey willbe archived and made available for usein the future. Take advice from nationalorganisations such as the ArchaeologicalData Service (see http://ads.ahds.ac.uk/ for contact details). Ask whoowns the collected data and thedelivered product.

l Finally, prepare a project brief, using astandard document as a base, such asthe English Heritage Metric SurveySpecification.

You may wish to carry out a small project first, before committing to a largersurvey, to help you fully understand thebenefits and limitations of the technique.Figure 20 describes a typical projectflowline. After identifying the need for a survey to be undertaken, a project brief should be established by the client.The project brief should includeinformation that helps the contractorunderstand the site-specific needs andrequirements of the survey. It should bewritten with direct reference to the survey specification, and should promptthe client for the relevant information.Once the project brief has been preparedit is put out to tender for surveycontractors to provide a method statementgiving details of how they intend toundertake the survey. The survey will thenbe commissioned and undertaken.

During this work the contractor should be guided by the method statement, butmay also want to refer to a standardspecification for guidance wherenecessary. Upon completion the client will use the project brief and standardspecification to undertake a qualityassurance (QA) check before acceptingthe survey and passing it into the archiveand/or on for use. Typically this is donethough a visual inspection of the data toensure that it shows what the user isexpecting. In other cases this QA processmight involve the comparison of thedelivered survey against check points.

3.2 Determining appropriate pointdensityOne of the key factors in commissioning a survey is being aware of what pointdensity and measurement accuracy isrequired to generate the level of‘deliverable’ you require in the project.Generally, using a point density of lessthan the quoted measurement accuracywill not provide useful information.For example, sampling every 1mm, whenthe measurement accuracy is 5mm.Based on standard mathematics used to determine appropriate minimumsampling intervals, and on the collectionof a regular grid of data, a simple guide to appropriate point densities isgiven in Table 2.

When preparing to commission a survey,a user should be aware of what is thesmallest sized feature he/she will requireto be detected. This may not be the same over the entire object/area of survey,so it may be appropriate to employdifferent point densities in different areas.The scanner used should have ameasurement accuracy of at least thepoint density of the scanning device used.(For example, a laser scanning with agiven accuracy of ±5mm should not beused to collect data at a point density of less than 5mm.)

3.3 Finding a contractorProfessional organisations, such as theRoyal Institution of Chartered Surveyors(RICS), trade organisations such as theSurvey Association (TSA) or staff withinEnglish Heritage will be able to help youto find an appropriate contractor.Alternatively, contact other projects,individuals or other organisations and askfor recommendations.

3.4 Laser safetyLaser light, in some cases, can be harmful.To enable users to determine the potentialrisk, all lasers and devices that use lasersare labelled with a classification,depending on the wavelength and powerof the energy the laser produces.Lasers used in survey applications mayhave risks associated with eye damage.The European Standard “Safety of LaserProducts – Part 1: Equipmentclassification, requirements and usersguide” (IEC 60825-1: 2001) providesinformation on laser classes andprecautions. It outlines seven classes oflasers:

l Class 1 lasers are safe under reasonablyforeseeable conditions of operation,including the use of optical instrumentsfor intrabeam viewing.

l Class 1M lasers are safe underreasonably foreseeable conditions ofoperation, but may be hazardous ifoptics are employed within the beam.

l Class 2 lasers normally evoke a blinkreflex that protects the eye, this reactionis expected to provide adequateprotection under reasonably foreseeableconditions, including the use of opticalinstruments for intrabeam viewing.

l Class 2M lasers normally evoke a blinkreflex that protects the eye, this reactionis expected to provide adequateprotection under reasonably foreseeableconditions. However, viewing of theoutput may be more hazardous if theuser employs optics within the beam.

10

Fig 20 The survey flow line.

Archive Use

Quality assurance

Survey delivery

Survey undertaken

Survey commissioned

Contractors prepare andsubmit method statement

(including costs)

Tender period

Survey required - prepareproject brief

SpecificationTable 2 Appropriate point densities (sampling resolutions) for various sizes of cultural heritage feature.

point density required to point density required tofeature size example feature give 66% probability that give a 95% probability that

the feature will be visible the feature will be visible

10000mm large earth work 3500mm 500mm1000mm small earth work/ditch 350mm 50mm100mm large stone masonry 35mm 5mm10mm flint galleting/large tool marks 3.5mm 0.5mm1mm Weathered masonry 0.35mm 0.05mm

l Class 3R lasers are potentiallyhazardous where direct intrabeamviewing is involved, although the risk islower than that for Class 3B lasers.

l Class 3B lasers are normally hazardouswhen direct intrabeam exposure occurs,although viewing diffuse reflections isnormally safe. This class of laser isgenerally not suited for surveyapplications.

l Class 4 lasers will cause eye or skindamage if viewed directly. Lasers of thisclass are also capable of producinghazardous reflections. This class of laseris not suited for survey applications.

Users of laser scanning systems shouldalways be aware of the class of theirinstrument. In particular the user shouldensure that the correct classificationsystem is being used. Refer to the IECstandard for more information on lasersafety: IEC 60825-1 2001, Safety of LaserProducts – Part 1: Equipment Classification,Requirements and User’s Guide.

Particular precautions and procedures areoutlined in the IEC standard for Class1M, Class 2M and Class 3R laserproducts used in surveying, alignment andlevelling. Those precautions, withrelevance to laser scanning are:

l Only qualified and trained personsshould be assigned to install, adjust andoperate the laser equipment.

l Areas where these lasers are usedshould be posted with an appropriatelaser warning sign.

l Precautions should be taken to ensurethat persons do not look into the beam(prolonged intrabeam viewing can behazardous). Direct viewing of the beamthrough optical instruments (theodolites,etc) may also be hazardous.

l Precautions should be taken to ensurethat the laser beam is notunintentionally directed at mirrorlike(specular) surfaces.

l When not in use the laser should bestored in a location where unauthorizedpersonnel cannot gain access.

3.5 Archived data sourcesIn some cases you may be able to usearchived data from commercialorganisations or government agencies,especially for airborne laser scanning oflandscapes and sites. However, be awarethat this data may have artefacts in thedata owing to processing, which may besignificant when performing analysis (seeCase Study 14). Also note that the point

density and measurement accuracy of thisdata may also not be sufficient for theanalysis required. Also consider the archiveissues for using such data (Fig 21).

4 From point cloud to usefulinformation

4.1 Typical workflowsThe commissioning of the survey is onlythe start of the survey process (see Fig 22for a general example with examples

of deliverable data). In order to turn scan data into a useful product the scans must first be registered, generallywith the use of external surveymeasurements, to provide some control.This will be done by the contractor,who will then, most likely, generate some defined deliverable output. At thisstage the user who has commissioned/procured the survey will want toundertake some form of analysis to help answer the questions that wereoriginally posed.

11

Fig 21 Elevation-shaded airborne laser scanning data (blue: low elevation; red: high elevation) for an urban area (datacourtesy of the Environment Agency; image courtesy of Newcastle University).

Fig 22 A typicalprocessing workflow.

Survey measurments Scanning on site

Scan registration

Analysis

Conclusions

Deliverable generation – for example

Point cloud Unrefined mesh

Rendered images 2D/3D drawings

Animations Decimated/edited mesh

4.2 Cloud alignment/registrationFor anything other than the simplestobject, a number of separate scans fromdifferent locations are usually required toensure full coverage of the object,structure or site. When collected, scans arebased on an arbitrary co-ordinate system,so to use several scans together theirposition and orientation must be changedso that each scan uses a common co-ordinate system (this may be based on thelocal site grid).

This process is known as cloud alignment,or registration. For example, scan one andscan two in Figure 23 and Figure 24 areinitially in separate reference systems andcannot be used together until they haveundergone a registration process, asshown in Figure 25. If the collected dataneeds to be referenced to a real world co-ordinate system, then it will be necessaryto provide external survey measurements.

In the case of airborne laser scanning thisis accomplished directly through the useof position and orientation observations.When using an arm-mounted triangula-tion laser scanner, co-ordinate measure-ments are collected in a known system,and so registration may not be required.

4.3 ModellingThe general term for the process required toturn the collected point cloud informationinto a more useful product is modelling, or,more descriptively, surface or geometricmodelling.There are a number ofapproaches that could be used to turn thepoint cloud into useful information.

For a small artefact or any object scannedwith a high accuracy triangulationscanner, the most typical product wouldbe a digital model of the object’sgeometry, probably in the form of ameshed model, such as a triangularirregular network (TIN). Figures 26 and27 show a point cloud before and aftermeshing, to form a TIN. In order togenerate a complete model of the subjectit is likely that some editing of the TINwill be required to fill holes where no datawas collected. The resulting TIN issuitable for use in several types of analysis.Figure 28 shows the result of meshingpoint data collected by laser scanning.

12

Fig 23 Scan one of the doorway.

Fig 24 Scan two of the doorway.

Fig 25 The scans of the doorway registered onto the sameco-ordinate system.

Fig 26 An unorganised point cloud prior to meshing showing a portion of the Upton Bishop fragment(courtesy of Conservation Technologies, National Museums Liverpool).

Fig 27 A meshed point cloud showing a portion of the Upton Bishop fragment (courtesy of ConservationTechnologies, National Museums Liverpool).

The ground points can then be used to generate a DTM, interpolating wherenecessary underneath buildings andvegetation.

13

Fig 28 An original stone fragment and a reconstructed geometric model from laser scanning data (courtesy of Conservation Technologies, National Museums Liverpool).

Fig 29 A drawing generated from laser scanning data and narrative imagery (courtesy of Tony Rogers, APR).

The options for processing a ground-based system are typically more varied.While a meshed model might be required,plans, profiles and sections (line drawings) could be generated by using the point cloud as a base from whichfeatures are traced, based on the edges inthe geometry and intensity data (Fig 29).However, this is not an automatic processand requires skill and experience on thepart of the users. The resulting drawing,without the underlying point cloud, willbe a fraction of the file size of the originaldataset.

With airborne laser scanning the most typical product is a digital terrain model (DTM).

The first task is to undertake aclassification on the available points.Using semi-automated algorithms thepoints that represent the ground can beidentified. The ground surface can beused as a reference to classify other pointsas ‘vegetation’ and ‘structure’ classes.

The DTM will initially be in the form of aTIN, where the surface is formed by aseries of interconnecting triangles. ThisTIN may also be used to create an inter-polated grid, in which each element in thegrid represents terrain surface elevation.A grid-based DTM might be more suitable for using within a mainstream GIS(Fig 21 is an example).

4.4 AnalysisThe delivery of a product derived fromlaser scanning data is only the start of the process of answering the originalresearch questions. Some form of analysis is likely to be required using thefinal product. In fact, some of this analysis may be best done during theprocessing stage itself. Considertalking/working with the contractor during the initial processing. Analysis,during or after the deliverable generationof data, should always includesupplementary data to support anyconclusions made. Consider howsupplementary datasets (such as historicmapping, or photos used within a GIS)might help (see Case Study 14).

As laser scanning provides three-dimensional data it lends itself very well to three-dimensional queries. Line-of-sight analysis allows a user to quantify if one part of the model can be seen from another location, eg 50% of the castle is visible from the valley floor.This procedure might be used in theanalysis of a landscape.

Another useful technique in analysing asurface is to use artificial raking light toilluminate a scene from directions notpossible by relying on sunlight alone (Fig 30; see Case Study 8).

Subtle features might also be identifiedusing vertical exaggeration. Byexaggerating the vertical scale at whichfeatures are displayed, slight variations in topography are often revealed.

This may be coupled with the use ofartificial raking light (Fig 31).

Neither of these analysis techniques wouldbe possible without detailed three-dimensional information, to which laser

scanning has greatly improved access.While laser scanning explicitly providesgeometry, most time-of-flight laserscanners also provide a value thatindicates the strength of the returninglaser signal. This intensity data may be

14

Fig 31 Elevation data from airborne laser scanning in the Witham Valley (top) and displayed with ten-fold verticalexaggeration (bottom) revealing possible early field systems to left of image (courtesy of Simon Crutchley, English Heritage,LiDAR courtesy of Lincolnshire County Council – Source Environment Agency).

Fig 30 A triangulated model of rock carvings with artificialraking light (courtesy Paul Bryan, English Heritage).

useful as an additional information sourceduring analysis, for example in theidentification of different stratigraphy in alaser scan of exposed soil. As mostscanners operate outside of the spectrumvisible to the human eye the intensityinformation collected is often slightlydifferent to that which is seen in reality.Such information can be useful, in somecases in differentiating between slightchanges in surface or material type.Figure 32 shows an example of how theintensity information from a scan of anarchaeological excavation can becompared with the record made onsite.

Three-dimensional geometric models mayalso be used to generate high-quality stillor animated scenes. Movies have beenused successfully to present what wouldotherwise be very large data quantitiesrequiring specialist viewing software andhardware. While such presentation doesnot provide an environment throughwhich a user can navigate freely, it doesserve a useful purpose in presenting anobject, site or landscape to a non-specialist group. Such models generallyinclude the use of image textures. Texturalinformation can often help to replicategeometric detail, and reduce the need forsome vertices.

5 Managing data

5.1 Reprocessing dataData is generated at a number of stagesduring a laser scanning survey. In order tobe able reprocess data at a later date auser should ensure that the mostappropriate data is available. Thefollowing diagram (Fig 33) summarisesthese stages and the data they produce:

Raw observations(As collected by the scanner)

t

Raw XYZ(As determined by the scanner)

t

Aligned XYZ(Determined by processing software/process)

t

Processed model(As chosen by the user)

Fig 33 Types of data arising from laser scanning.

Raw observations are not universallyavailable, and data formats differ betweenmanufacturers. Raw XYZ data is, instead,the most preferred data source forreprocessing, which could include taskssuch as realignment of scans. Whateverdata you have, you should also ensure thatyou have a record of the processinghistory, including information on any re-sampling (often referred to as ‘decimation’when used in reference to datamanipulation) of the data.

If you want to ensure that data can beused in the future, it is recommended thatservice providers should retain theproprietary observations after completionof the survey for a minimum of six years.This should include: field notes and/ordiagrams generated while on site; the rawand processed data used for the finalcomputation of co-ordinate and levelvalues; and a working digital copy of themetric survey data that forms each survey.

5.2 Data formats and archivingData exchange formats are used to makethe transfer of data between users easier.Proprietary formats should be avoided for

this purpose. A simple text file (oftenreferred to as ASCII) providing fields forXYZ co-ordinates, intensity informationand possibly colour (RGB) informationwould generally be sufficient for thetransfer of raw data between one softwarepackage and the next. However, in orderto standardise the transfer of suchinformation, and ensure that importantinformation is not lost in transfer, it mightbe appropriate to consider a formal dataexchange format. An emerging transferformat for point cloud data is the LASformat, overseen by the American Societyfor Photogrammetry and Remote Sensing(ASPRS). This open source format wasoriginally developed for the transfer ofairborne laser scanning betweencontractors and packages. However, it canalso be used to transfer ground basedlaser scanning data. The LAS format iscurrently being revised by its steeringcommittee to Version 2.

Of perhaps more concern to the end userare the formats chosen to deliver theactual product to be used. Obviously theformat needs to be compatible with thetools you intend to use. A good generalpurpose format for the delivery of meshedmodels is the Alias Wavefront OBJ format.

The type of ‘deliverable’ will dictate therange of data formats that can be used.For typical raw and interpreted scan datathe following delivery formats should beconsidered:

l Digital Terrain Models (DTM): any textbased grid format

l TIN models: Wavefront OBJl CAD drawings: DXF, DWGl movies/animations: QuickTime MOV,

Windows AVIl rendered images: TIFF, JPGl replication: STL

The deliverable product may also includewritten reports, which should generally be

15

Fig 32 A laser scan of an excavated section with intensityinformation (left), and a stratigraphic record of the ditch madeby an archaeologist on site (below) (courtesy of NewcastleUniversity and Archaeological Services University of Durham).

delivered in PDF format fordissemination, and with an ASCII text fileversion also provided for archiving.For detailed guidelines on issues ofarchiving, including appropriate fileformats, readers should refer to theArchaeological Data Service’s (ADS)project ‘Big Data’.

5.3 MetadataAn important component of the datamanagement process is the definition andmanagement of metadata: data about thedata. This is especially true whensubmitting the final record to archivingorganisations such as the ADS. The veryminimum level of information that mightbe maintained for raw scan data mightinclude the following:

l file name of the raw datal date of capturel scanning system used (with

manufacturers serial number)l company namel monument namel monument number (if known)l survey number (if known)l scan number (unique scan number for

this survey)l total number of pointsl point density on the object (with

reference range)l weather conditions during scanning

(outdoor scanning only)

For full details of the metadata requiredby English Heritage, see “An Addendumto Metric Survey Specifications forEnglish Heritage – the Collection andArchiving of Point Cloud Data Obtainedby Terrestrial Laser Scanning or otherMethods”. This document is currentlyavailable as a pdf file download from theHeritage3D website www.heritage3d.org, but will soon be inserted into the2007 revision of the “Metric SurveySpecifications of English Heritage” (ISBN1 873592 57 4, published by EnglishHeritage 2000; reprinted March 2003).

6 Helping you to decideAsking yourself the following questionswill help you to better understand whatyour requirements are and whether laserscanning, in its various forms, is suitable.It will also help to identify possiblealternatives.

6.1What outputs are wanted?Scanning can contribute to a whole rangeof outputs, so deciding what outputs you

require will help you to determine anappropriate project brief. Outputs mightinclude a highly edited surface mesh, two-dimensional drawings, rendered movies oreven virtual environments. Other forms ofdata, such as images and survey control,are likely to be required to contribute tothese outputs.

The scale of your output is a key decision,which will help determine the accuracy ofyour product and the required pointdensity. Next, think about how you willuse the output. Does it need to be hardcopy, perhaps for annotation on site? Doyou need to be able to edit it yourself,view it as part of some interpretationactivity or will it simply be used fordissemination and reporting, for exampleas part of a presentation? If there areother potential users of the output, forexample within a project team, considerwhat sort of output they might require.

6.2 How big is the subject?The size of the object or site in questionhelps to define the type of laser scanningthat would be appropriate to apply. Atriangulation laser scanner could providemeasurements to an accuracy of less than1mm and point densities of around thesame scale, so would be ideal for therecording of a small artefact or statue. Afeature on a building, although larger,might also be suitable for measurementusing a triangulation scanner, although ifthe object is fixed in place, access to itshould be considered. Alternatively, itmight be suitable to use a system basedon time-of-flight measurement.

At the scale of a building façade or of anentire building, measured survey usingtriangulation scanners would take anunjustifiably long time and would providedata at far too high a resolution (in additionto being very expensive).Therefore, giventheir suitability for larger objects, owing totheir greater working range, a time-of-flightscanner would be more appropriate.

For entire sites, where the topography ofthe site is of interest, time-of-flightscanning, using a scanner with a 360-degree field of view would be feasible,whereas for an entire landscape,incorporating a number of sites ofinterest, airborne survey would be theonly likely solution.

6.3 What level of accuracy is required?This is typically related to object size andthe purpose of the survey. A common

answer is ‘the best that you can do’, butthis is not always helpful in deciding whattype of technique should be used. It isperhaps more correct to ask what is theoptimum accuracy that balances the needsof the task, the capability of the techniqueand the budget available.

6.4 What resolution of measurement?Again, this is typically related to objectsize and purpose of survey. Resolution isthe density of co-ordinate measurementsover the subject area. With a subject thathas a complex shape or sharp edges, it isnecessary to have high-resolutionmeasurements so that the resulting datahas a high fidelity to the original subject.There might be situations where the bestoption is to combine a number ofresolutions. Low-point density in areas ofreduced complexity, or where high levelsof detail are not required, along withhigher resolution areas of high complexityand interest. For example, the recordingof a building façade may require veryhigh-resolution measurements of smallcarvings and tympana while, incomparison, the rest of the buildingrequires a basic record of dimensions andlayout. The choice of resolution shouldalso be balanced against the accuracy ofthe system measurements.

6.5 Does the survey need to be geo-referenced?When working on structures, buildings,sites and landscapes it is likely that thedata will need to be linked to a local ornational reference system. This makes ifpossible to use the collected dataalongside other spatial datasets on thesame system. It is less likely that a smallobject or feature will need to bereferenced to a common system, althoughits original spatial location and orientationmight need to be recorded.

6.6 Time and access restrictions?Access and time might be unlimited. Forexample, the object might be brought to astudio-based scanner. Alternatively, accessto the subject may be easy, perhapsbecause temporary scaffolding is in place,but time may be restricted because thescaffolding will be dismantled, makingfuture access impossible unless newscaffolding is erected. Note that whilescanning from a static position requires astable platform, scanning from scaffoldingor from a hydraulic mast or cherry pickeris possible, although care should be takento ensure that the scanner remains stableduring operation (Fig 34).

16

Access might be restricted on health andsafety grounds, because a building isunsafe, making a survey only possiblefrom a few locations. In an archaeologicalexcavation, survey may be time-critical, asrecording is required at each part of theexcavation and cannot be repeated. Thisrequires scanning to be available on siteduring excavation.

The weather can also impose limitations.Scanning in heavy rain is generallyunsuitable, as rain on the scan window canrefract the measurement beam. Airbornesurvey is, to some extent, also restricted byweather. Survey might also be required ata particular time, for example if datacollection is required when trees are in leafor when bare (in surveying terminology‘leaf on’ or ‘leaf off ’ conditions).

6.7 Is three-dimensional informationrequired?If yes, consider how the information isgoing to be used. This will help you or thecontractor to determine the processingthat will be required on the laser scanningdata. Even if the answer is no, and youonly need two-dimensional measurementsand dimensions, laser scanning may stillbe useful. Laser scanning can be used toprovide line drawings in section, profilesand plans. It is especially useful whenaccess to a site makes it difficult to useconventional methods (see Case Study11). The way in which laser scanning

enables direct integration ofthe collected data on site canalso help a contractor reducethe likelihood for revisits.

6.8 BudgetAlthough laser scanning isgenerally regarded as a high-cost technique, it can bejustified, as the informationrequired may not be availablein any other way. If thebudget is limited, or non-existent, laser scanningprobably is not a techniquethat you can use. Where it isused, it is advisable to try toensure that it can be used inmany different ways, so as toprovide best possible valuefrom its commissioning.

6.9 Can you do thisyourself?

It may be possible to undertake the datacollection and data processing yourself.However, scanning requires specialistskills in order to achieve a precise andreliable product. This might include skillsin providing precise survey controlmeasurement and/or specialist skills in 3DCAD or GIS. If this is your first project,using a contractor is advisable.

6.10 What are the alternatives?Digital photogrammetry is the techniqueto which laser scanning is most compared.Photogrammetry is increasingly easytoday, compared to 5–10 years ago whenit generally required the use of specialist

analytical instruments (Fig 35). It canprovide a highly scalable and accuratemethod of measuring surface topography.It can also be used from the air, or fromthe ground, although as a non-activemeasurement technique (photographsonly record the light reflected from thesun or other illumination source) it is lessable to measure through small gaps inforest canopies. Thus, where a site iscovered in woodland, laser scanning maybe the only solution that can providemeasurement to the forest floor.

A terrestrial, topographic survey usingdifferential GPS (Fig 36) or a reflectorlesstotal station survey might provide a lower-cost site survey, but over a large landscapethis might not be suitable. Terrestrialsurvey using reflectorless EDMmeasurement can also be used to generatebuilding façade elevations, in real time orusing post processing in CAD. Handrecording using tape and plumb line canprovide accurate records of small features,objects or structures.

17

Fig 34 Laser scanning from scaffolding (courtesy of DuncanLees, Plowman Craven Associates).

Fig 35 An operator using a digital photogrammetricworkstation (courtesy of English Heritage).

Fig 36 A digital terrain model generated by ground basedGPS survey (courtesy of English Heritage).

7 Where to find out more

7.1 Charters and guidanceThe aims of the recording within thescope of conservation and restoration areprovided in the Venice Charter, drawn upin May 1964 (see http://www.international.icomos.org/e_venice.htm).

Overall guidance and a detailedspecification for the use of recordingtechniques are available from the English Heritage Metric SurveySpecification. Contact the EnglishHeritage Metric Survey Team for more information (contact details aregiven below).

7.2 Organisations There are a number of organisationswhose members have expertise in laserscanning and in the provision of three-dimensional survey. They may be able to help you find an appropriatecontractor, or be willing to talk over your particular needs.

The Archaeological Data ServiceDepartment of ArchaeologyUniversity of YorkKing’s ManorExhibition SquareYork YO1 7EPhttp://ads.ahds.ac.uk/

English Heritage Metric Survey Team, YorkEnglish Heritage37 Tanner RowEnglish HeritageYork YO1 6WPhttp://www.english-heritage.org.uk/server/show/conWebDoc.4143

Remote Sensing and Photogrammetry Society (LaserScanning and LiDAR Special Interest Group)c/o Department of GeographyThe University of NottinghamUniversity ParkNottingham NG7 2RDUnited Kingdomhttp://www.rspsoc.org/

Royal Institute of charted Surveyors(RICS) Mapping and PositioningPractice Panel12 Great George StreetParliament SquareLondon SW1P 3ADUnited Kingdomhttp://www.rics.org/

The Survey AssociationNorthgate Business Centre38 NorthgateNewark-on-TrentNotts NG24 1EZUnited Kingdomhttp://www.tsa-uk.org.uk/

7.3 BooksTo date, there are no books thatspecifically cover the use of laser scanning in cultural heritage. However,there are some useful guides to the needs and methods of measured survey in cultural heritage. These booksmake some reference to the use of laser scanning:

English Heritage 2003 Measured andDrawn – Techniques and practice for themetric survey of historic buildings, 62 pages

Dallas, R W A 2003 A guide forpractitioners – Measured survey and buildingrecording, Historic Scotland, 180 pages

Böhler, W and Marbs, A 2004 Acomparison of 3D scanning andphotogrammetry for geometric documentationin cultural heritage. Fachhochschule Mainz, 86 pages (in German: Vergleich von 3D-Scaning und Photogrammetrie zurgeometrischen documentation imDenkmalbereich).

7.4 Journals and conference proceedingsThere is no specific journal for laserscanning, but many major journals thatcover survey techniques and culturalheritage have, in the past, included paperson the subject:

ISPRS Journal of Photogrammetry andRemote Sensing (Amsterdam: Elsevier)

The Photogrammetric Record (Oxford:Blackwell)

Journal of Architectural Conservation(Dorset: Donhead)

There is also a range of professionaljournals that often provide annualsoftware and hardware reviews on laserscanning:

Geomatics World (PV Publications, UK)

Engineering Surveying Showcase (PVPublications, UK)

GIM International (GITC bv,Netherlands)

There are also a number of regularconferences where research on, and theapplication of, laser scanning is presented,and who publish comprehensiveproceedings:

Symposia for the International Committee for Architectural Photogrammetry (CIPA).Held every two years, the Proceedings ofthis symposium can be found online at:http://cipa.icomos.org/index.php?id=20

International Archives of Photogrammetryand Remote Sensing. Proceedings for the main congress (held every four years) and for the mid-term symposiums(held once in the four years betweencongresses) can be found at:http://www.isprs.org/publications/archives.html

7.5 WebsitesAt the time of writing the followingwebsites provide useful information anddetails of projects and free software:

Heritage3D project – Information and guidance on the use of laser scanning in cultural heritage,http://www.heritage3d.org

The English Heritage Big Dataproject at the Archaeological DataService – Guidelines on archiving of archaeological data and lists of software packages (including free dataviewers), http://ads.ahds.ac.uk/project/bigdata/

The English Heritage aerial surveyand investigation team – Informationon the work of English Heritage’saerial survey team, including theirexperience of LiDAR,http://www.english-heritage.org.uk/aerialsurvey

Stonehenge laser scanning – Anexample of laser scanning at one ofEnglish Heritage’s most well known sites,http://www.stonehengelaserscan.org/

i3Mainz – A good source of technicalinformation and case studies on laserscanning equipment and application,http://www.scanning.fh-mainz.de/

7.6 Training Manufacturers of laser scanningequipment and software will be pleased to provide training. Other organisationsthat may be able to provide sources oftraining includes university departmentsand commercial survey companies.

18

8 Glossary

3D Having three-dimensions,characterized by Cartesian (XYZ) co-ordinates

airborne laser scanning The use of alaser scanning device from an airborneplatform to record the topography ofthe surface of the earth

ADS Archaeological Data Service,University of York

CAD Computer aided designCIPA International Committee for

Architectural Photogrammetrycultural heritage Refers to tangible and

intangible evidence of human activityincluding artefacts, monuments, groupsof buildings and sites of heritage value,constituting the historic or builtenvironment

data voids Sections within the pointcloud, more than twice the point densityof the scan in size, which contain nodata despite surface information on theobject itself

DEM Digital elevation model, atopographic model of the bare earththat can be manipulated by computerprograms and stored in a grid format

DSM Digital surface model a topographicmodel of the earth’s surface (includingterrain cover such as buildings andvegetation) that can be manipulated bycomputer programs

DTM Digital terrain model, a topographicmodel of the bare earth that can bemanipulated by computer programs

geometric accuracy The closeness of ameasurement to its true value. It iscommonly described by the RMS error

geometric precision The distribution of a set of measurements about theaverage value, which is commonlydescribed by the standard deviation.All reference to the standard deviationof a quantity should be accompanied bythe probable error value eg ±3 mm(67% probable error) – sometimesreferred to as repeatability.

GIS Geographical information systemGPS The global positioning system – a

US satellite positioning system used toposition an aircraft during an airbornesurvey, or used as a ground basedsurvey technique

LAS Abbreviation for data format – .laslaser Light Amplification by Stimulated

Emission of Radiation: an electronic-optical device that emits coherent lightradiation

laser scanning From a user’s point of view, a 3D scanner is any device that collects 3D co-ordinates of a given region of an object surfaceautomatically and in a systematicpattern at a high rate (hundreds orthousands of points per second)achieving the results (ie 3D co-ordinates) in (near) real time.

LiDAR Light Detection and Ranging –often used to refer to airborne laserscanning but can also apply to someground based systems

mesh A polygonal subdivision of thesurface of a geometric model

metadata Data that is used to describeother data and a vital component of thedata management process

model An expression that should bequalified by the type of model, eggeometric model. A geometric model is,typically, a representation of a three-dimensional shape.

peripheral data Additional scan datacollected during the scanning process,but not explicitly defined in the project brief

point cloud A collection of XYZ co-ordinates in a common co-ordinatesystem that portrays to the viewer an understanding of the spatialdistribution of a subject. It may alsoinclude additional information, such as an intensity or RGB value.Generally a point cloud contains arelatively large number of co-ordinatesin comparison with the volume thecloud occupies, rather than a few widelydistributed points.

point density The average distance betweenXYZ co-ordinates in a point cloud

recording The capture of informationthat describes the physicalconfiguration, condition and use ofmonuments, groups of buildings andsites, at points in time. It is an essentialpart of the conservation process (see theVenice Charter – International Charterfor the Conservation and Restoration ofMonuments and Sites, May 1964).

registration The process of transformingpoint clouds onto a common co-ordinate system

repeatability Geometric precision (seeabove)

scan orientation The approximatedirection in which the scan is made ifthe system does not provide a 360-degree field of view

scan origin The origin of the arbitraryco-ordinate system in which scans areperformed. When the scan origin istransformed into the site co-ordinatesystem it becomes the scan position.

scan position The location, in a knownco-ordinate system, from which a singlescan is performed. If the system doesnot perform a full 360-degree scan,several scans may be taken from thesame scan position, but with differentscan orientations.

scanning artefacts Irregularities within ascan scene that are a result of thescanning process rather than features onsubject itself

survey control Points of known locationthat define a local reference frame inwhich all other measurements can bereferenced

system resolution The smallestdiscernable unit of measurement of thelaser scanning system

terrestrial laser scanner Any ground-based device that uses a laser tomeasure the three-dimensional co-ordinates of a given region of an objectssurface automatically, in a systematicorder at a high rate in (near) real time

TIN Triangulated Irregular Network

19

IntroductionIn addition to photographic documentation, Upton BishopParochial Church Council wanted an accurate three-dimensionalrecord of the sandstone fragment shown below.The fragmentmeasures approximately 400mm x 200mm x 210mm andbelongs to the Parish of St. John the Baptist, Upton Bishop.

Upton Bishop sandstone fragment.

Instruments and softwareA ModelMaker X laser scanning system with a 70mm stripewidth, mounted on a 7-axes Faro gold arm was used for datacapture. Sensor-object separation was maintained at 50mmthroughout. Sensor and arm calibration had an average RMS error of 0.03mm. Scanning was carried out in ourstudios in Liverpool. Four scanning stations were required to capture the whole object. There was no sampling duringdata capture. The software used to collect the data was 3D Scanners UK ModelMaker V7 beta release. Oncescanning was complete, the data was 2D sampled at u = 0.2and v = 0.2 mm using MM V7. Polyworks V8 (InnovmetricSoftware, Inc) was employed for data alignment, mergingand post-processing. The maximum edge length parameterused during meshing was 0.2mm. Rapidform2004 PP2(INUS Technologies, Inc.) was used for registering andmerging the data from the four scanning stations together.The average maximum deviation between the data from eachof the scanning stations was 0.04mm. Any areas where theFaro arm had been at full stretch during scanning weredeleted, and data from another station was substituted.

Abnormal faces were deleted and all holes were filledmanually using Rapidform2004.

Production of 3D fly-throughs in AVI format wasundertaken in 3D Studio Max (AutoDesk Media andEntertainment). Photographic documentation was capturedusing a Minolta Dimage 5 3.3 megapixel digital camera at aresolution of 1600 x 1200 pixels, mounted on a tripod. Eachimage was manually white-balanced.

Why was scanning selected?Upton Bishop Parochial Church Council wished to have a3D archive of this important object. The Church Councilwishes to improve access to this delicate and importantobject, while limiting handling of the piece to a minimum.Upton Bishop Parochial Church Council and otheracademics are attempting to learn more about the fragment;currently it is undecided whether it is Early Christian orRoman in origin. The fragment is very fragile; any handlingresults in some surface loss. In addition to the photographicdocumentation, the 3D digital model can be supplied as a setof screenshots, a fly-through or as a virtual model to a varietyof interested parties worldwide. In this way the fragment canbe made available to a large number of scholars and membersof the public, while limiting any potential damage to thisimportant object. A 3D virtual model enables Upton BishopParochial Church Council to provide the archive, withphotographic documentation, to experts and interestedparties to help establish the provenance of the piece. Inaddition, the possibility exists to create a replica of thefragment if the original is ever stolen, damaged or destroyed.An extremely accurate replica (either to scale, or to differentsizes) can be made in the original material, sandstone (oralternatively, synthetic materials). The data obtained by laserscanning is used to control the tool path of a computernumerically controlled (CNC) milling machining.

What problems were encountered?Owing to the fragile and friable nature of the surface of thefragment, handling and positioning of the piece was kept to aminimum. As the piece could not be propped up, and had toremain horizontal throughout scanning, a larger number ofscanning stations than usual was required for a piece of thissize. As scanning was undertaken in the studio, the datacaptured was exceedingly accurate. Registration and mergingthe separate stations did not pose any problems.

What were the final ‘deliverables’?Upton Bishop Parochial Church Council were supplied witha copy of the raw scan data (in SAB2, a 3D Scanners file,and ASCII format), as well as the completed post-processeddata in STL format, photographic documentation of theoriginals, and metadata detailing how scanning and post-processing was undertaken. A fly-through where the objectspins slowly through 360° in AVI format was also provided.

All rights reserved.This case study is published with kind permission of the contributor.All images reproduced by kind permission of Upton Bishop Parochial Church Council.

C A S E S T U DY 1

Creating a 3D archive of the Upton Bishop fragmenttype: non-contact recording using laser scanningkeywords: non-contact, recording, 3D, archive, laser scanning, Upton Bishop, sandstone

20

Upton Bishop fragment: screenshots of 3D data.

21

IntroductionPrestbury Church (Cheshire) has an important Normandoorway. The doorway and surround measures approximatelyH 6m x W 2.8m x D 0.7m, the smallest detail to be recordedon the doorway and surround was approximately 5mm. Thesandstone surface is badly weathered and very friable. Weprovided Prestbury Parochial Church Council with a full dataset. The aim of the work was to provide an accurate 3D recordof the doorway at the time of recording.

Instruments and softwareA Mensi S25 laser scanner mounted on a large tripod wasused for data capture. The Mensi S25 works bytriangulation. Sensor-object separation ranged between 3mand 7m throughout data capture. The S25 records points

across the surface in a grid in anautomated fashion, once the user hasdetermined the scan area and grid size.The average grid size used for recordingthe doorway was 3mm. Detailed sectionswere captured at 2mm, areas of lessdetail, such as the door and thesurrounding brickwork, were recordedwith a maximum grid size of 4.2mm.

The resolution of the data isdependent on the grid size determinedby the operator and the accuracy of thesystem. The Mensi S25 has a relative

accuracy of 0.619mm at an object-scanner separation of 5m, with an error of ±1.2mm. During scanning at a distance of 3m, with a grid size of 2mm, the standarddeviation was recorded as 0.86mm. A calibration check wasperformed on the system prior to scanning and oncompletion of data capture, on a 999.96mm carbon fibre bar at a distance of 5m.

The measurements had an average error of 0.471mm.Data capture required 10 scanning stations and took 22 hourson site to complete. Power was supplied via a mains socketwithin the chapel. The scanner, associated equipment andoperators were housed in a tent to protect them from theelements. Scanning could only take place in low lightingconditions or complete darkness. Recording took place inDecember, when the evenings are long. The data was recordedusing Scanworks software (Mensi-Trimble). Polyworks V8(Innovmetric Software, Inc.) was employed for data alignmentof scanning patches and meshing. Registration and merging ofthe different stations and post-processing was undertaken inRapidform 2004 SP2 (InusTechnologies, Inc).

The average shell-shell deviation during registration was0.65mm. A small amount of manual hole filling and localised smoothing was required. In total, post-processing of the data took 35 hours. Production of 3D fly-throughs inAVI format was undertaken in 3D Studio Max (AutoDeskMedia and Entertainment). Photographic documentationwas captured using a Minolta Dimage 5 3.3 megapixeldigital camera at a resolution of 1600 x 1200 pixels, mountedon a tripod.

Why was scanning selected?We wished to examine the use of an S25 laser scanning systemfor the recording of an outdoor architectural feature. We werepleased with the data we obtained. We felt it was of goodresolution and accuracy with regard to the size of the scanningarea and the level of detail on the doorway and the surround.

What problems were encountered?The S25 requires low levels of lighting to be able to capturesufficient data to record the surface. For this reason, ratherthan erect a very large and expensive scaffold housingaround the entire scanning areas, scanning was undertakenat dusk and nighttime. The laser beam used in the S25 is in aclass of laser (class 3A, according to the CDRH 21 CFR1040 standard) that can cause damage to the eye, if one wereto look directly into the beam. When the system is scanning,the beam is constantly moving and the blink reflex willprotect the eye from damage. It is imperative that no onelooks directly into this beam. The area in which scanningtakes place was sectioned off from public access using conesand hazard warning tape. In addition, highly visible warningsigns were used around the site. Despite the system beingsemi-automated, the equipment was never left alone. Oneoperator was always present to ensure no one entered thescanning area, and to monitor data capture. If needed, thesystem has an emergency stop button that will pause datacapture and shut off the laser beam. This can also be usefulwhen wildlife gets in the way of data capture. Our teamalways sends two operators onto site with this piece ofequipment for these reasons.

What were the final ‘deliverables’?Prestbury Church Parochial Council were supplied with acopy of the raw scan data (in SOISIC (Mensi S 25 file andASCII format), as well as the completed post-processed datain STL format, photographic documentation of the originaland an AVI flythrough of the doorway.

C A S E S T U DY 2

Recording a Norman doorway, Prestbury Churchtype: triangulationkeywords: non-contact, laser scanning, doorway, architectural fragment, documentation, recording

The Norman doorway: detail, 3D screenshot of data.

All rights reserved.This case study is published with kind permission of the contributor.

IntroductionAs part of the English Heritage Rock Art Pilot Project(1999), laser scanning as a method of documenting rock artin the field, was examined. The petroglyphs studied arelocated on Rombald’s Moor in West Yorkshire. The areasscanned were approximately 1.2m x 0.5m in size. Laserscanning was found to be a good technique for the 3D non-contact recording of rock art. The equipment is suitable foruse in the field – even in remote locations.

The petroglyphs on Rombald’s Moor,West Yorkshire.

The data obtained documented the petroglyphs to a highlevel of detail. Importantly, the results were not subjected tolighting conditions at the time of data capture. Indeed, oncethe data had been post-processed, and was examined undervarying lighting conditions, a distinct wear pattern waslocated on the surface of one of the rocks. This pattern hadnot been discernable from photography, nor with the nakedeye. The results of laser scanning can be exploited in a widevariety of imaging formats, providing a flexible digitalarchive. Images of contour maps of the surfaces and a scalereplica in polyurethane were also produced from the data.

Instruments and softwareA ModelMaker H laser scanner (3D Scanners UK) mountedon a six-axes Faro silver arm was used to record thepetroglyphs. A sensor with a 40mm stripe width was used,and the scanner recorded at a rate of twelve stripes per

second. Stripe-stripe separation is dependent on the speed atwhich the sensor is moved across the surface by the operator,and ranges from 4mm–0.2mm.

The petroglyphs were recorded with high accuracy, hencewith a slow scanner speed, producing a dense point cloud.Data was captured using ModelMaker V1 software. Meshingand post-processing was also carried out in ModelMakersoftware. All data was checked on site.

Areas that may have been missed during data capturewere easily highlighted and ensured as complete a data set aspossible was recorded. The raw point cloud was also crudelymeshed on site to check the quality of the scan data.ModelMaker V1 was used to create a contour map of theindividual rocks. Each rock art panel required 3–4 hours toscan, including the time it took to move the equipment andprepare it for data capture.

3D contour map of one of the rocks.

Why was scanning selected?Commonly used 2D techniques such as sketching, rubbing,and photography employed to record rock art are subject tolighting conditions. Moreover, they suffer inaccuracies owingto the difficulty of rendering a 3D surface in 2D. Close rangephotogrammetry, while providing a 3D archive, is still subjectto the lighting conditions at the time of recording. Mouldingand casting techniques, while both 3D and non-lightsubjective, can damage the weathered surface of a petroglpyh.

Non-contact recording using laser scanners provided asolution to all of these problems. In addition, by fixingdatum points close to the rock panel it is possible to rescanthe rock at a later date. The two data sets can then becompared to measure and monitor any changes or decay onthe surface of the petroglyph.

What problems were encountered?The ambient light levels on the moor were too bright forlaser scanning (scanning was undertaken during summer).As this was anticipated, a small tent was erected over the site.

C A S E S T U DY 3

3D recording of three petroglyphs on Rombald’s Moor,West Yorkshiretype: arm-mounted triangulation scannerkeywords: 3D recording, documentation, in-situ, laser scanning, rock art, petroglyphs, virtual images, digital archive,Rock Art Pilot Project, Rombald’s Moor

22

Laser scanning on Rombald’s Moor.

The tent suitably reduced light levels to enable data capture.Calibration of the equipment on site was difficult. There wasa lack of solid surfaces to mount the geometric cube used for

calibration. It was necessary to calibrate the arm off-site,move the equipment with care, and check the calibration oncompletion of scanning, off-site. In this way it was possibleto ensure that calibration had remained within specificationthroughout data capture.

What were the final ‘deliverables’?The raw data was provided in CTA (3D Scanners file) andASCII format. The post-processed data was provided as STLfiles, and the 3D contour maps were delivered in IGESformat. In addition, the data obtained by laser scanning wasused to produce a scale replica of one of the rocks inlightweight polyurethane using 3-axes computer numericallycontrolled (CNC) machining.

All rights reserved.This case study is published with kind permission of the contributor.

Screenshots of the 3D archive of the three petroglyphs

C A S E S T U DY 4

Non-contact 3D recording and replication of medieval Graffiti at the Tower of Londontype: arm-mounted triangulation scannerkeywords: non-contact, laser scanning, replication, CNC machining, exhibitions, Historic Royal Palaces, Medieval graffiti,Arundell, Dudley

IntroductionIn 2003 Historic Royal Palaces re-designed the exhibitionspace within the Beauchamp Tower at the Tower of London.The Medieval graffiti carvings in the upper room were to bereplicated to improve visitor access. Owing to the fragilesurface of these carvings traditional moulding techniquescould not be used.

The Dudley carving: Beauchamp Tower.

We employed non-contact 3D recording using laser scannersto create digital 3D models of the carvings. The dataproduced in this way can be used to control the tool path of

a robotic milling machining. From our scan data, scalereplicas were produced using 3-axis computer numericallycontrolled (CNC) machining. The replicas were installed inthe new exhibition in December 2003 and greatly improvevisitor access to these important carvings. The originalgraffiti remains on public display, behind sheets of glass forprotection, in the upstairs room of the Beauchamp Tower.Visitors can explore the graffiti inscriptions from the replicasprior to visiting the upstairs area, and are encouraged totouch the replicas.

Instruments and softwareA ModelMaker X laser scanning system with a 70mm stripewidth, mounted on a 7-axes Faro gold arm was used for datacapture. Sensor-object separation was maintained at 50mmthroughout. Arm calibration had an RMS error of 0.039mmand sensor calibration had an RMS error of 0.029mm.Sensor checks, both in the studio and on site at theBeauchamp Tower, had RMS error values of 0.04mm.One scanning station per carving was required and there was no sampling during data capture.

The software used to collect the data was3D Scanners UK ModelMaker V7 betarelease. Once scanning was complete, the datawas 2D sampled at u = 0.2 and v = 0.2 mmusing MM V7. Polyworks V8 (InnovmetricSoftware, Inc) was employed for dataalignment, merging and post-processing.Rapidform2004 PP2 (INUS Technologies,Inc) was used for post-processing andpreparation of the data for CNC machining.

The ModelMaker X laser scanning system.

23

24

Production of 3D flythroughs in AVI format wasundertaken in 3D Studio Max (AutoDesk Media andEntertainment). Photographic documentation was capturedusing a Minolta Dimage 5 3.3 megapixel digital camera at aresolution of 1600 x 1200 pixels, mounted on a tripod. Eachimage was manually white-balanced. The protective glasssheets over the graffiti were removed prior to data capture.

The Arundell carving: screenshot of 3D data.

Why was scanning selected?Owing to the friable nature of the surface of the carvings, theconservation department at Historic Royal Palaces had deemedthe carvings too delicate to take moulds from the originals.

What problems were encountered?We had to use a collapsible lightweight tripod due to thenarrow staircase that provided access to the scanning site. Inaddition, we were unable to attach or glue the tripod to thefloor in any way. The floor is too the delicate. Moreover, wewere unable to totally exclude movement of the floor.Despite the room being closed to members of the public forthe duration of data capture, we were not allowed to remainalone during scanning at this military site. Consequently, weobserved higher than normal movement during scanning,particularly between stripes. We were able to re-align thestripes using Polyworks to within 0.2mm.

The scan data of the Arundell carving exhibited someunusual features. Scanning features that are best visually

described as ‘pixilation’ along some of the sharp edges of the inscription were observed in the 3D model. This could not be explained by a lack of data. It appears to be due to the nature of the patination of the letters (the blackand white paint in the inscription), combined with thedirection of the stripe during scanning. Possible movement of the wooden floor while scanning may also have led toerrors in the data. The most pronounced example of this was observed on the A of Arundell.

What were the final ‘deliverables’?Scale replicas of each carving were milled in high-density cast resin (Ebablock 1200 f+f, Denaco UK) using three-axesCNC machining. Minimal hand finishing of the replicas by our sculpture conservators was required to removemachining marks and ‘pixelation’ in small areas on theArundell carving. Chisels, rifflers and scalpels were used.During patination of the replica graffiti, alkyd paints wereapplied using brushes, cloths and sponges. The surface was finished with a matt varnish.

In addition, Historic Royal Palaces were supplied with acopy of the raw scan data (in SAB2, a 3D Scanners file,and ASCII format), as well as the completed post-processeddata in STL format, photographic documentation of theoriginals and the replication process and metadata detailinghow scanning and post-processing was undertaken.

The replicas installed in the new exhibition space at theBeauchamp Tower,Tower of London.

All rights reserved.This case study is published with kind permission of the contributor and Historic Royal Palaces.

C A S E S T U DY 5

Copying Caligulatype: arm mounted triangulation scannerkeywords: non-contact, laser scanning, 3D virtual model, replication, replica, CNC-machining marble, polychromesculpture, Caligula, colour reconstruction

IntroductionIn the collections of the Ny Carlsberg Glyptotek (Copenhagen,Denmark) is a marble bust of the Emperor Caligula. Thesculpture was probably carved between 39 and 41 AD.Originally such sculptures were painted (polychrome) and thispiece has traces of the original polychromy remaining. Romanmarble sculptures retaining their original polychromy areexceedingly rare. The curators and conservators in Denmark

wished to study the pigments to determine their exactcomposition and then reconstruct a possible colour scheme ona replica object. Their intention was to display the original anda painted replica side-by-side. Due to the fragile pigmentedsurface of the bust, traditional moulding techniques could notbe used. We employed non-contact 3D recording using laserscanners to create a digital 3D model of the sculpture. Thedata produced in this way can be used to control the tool path

of the surface. The taking of a cast would in all probabilitydamage the pigmentation. In addition, this process wouldresult in a plaster or other synthetic material replica, not amarble copy. The curators of the project wanted to examine a colour reconstruction onto the original material – marble.A sculptor copy-carving the bust may have led to someelement of re-interpretation of the piece, no matter howminor or unintentional. For these reasons replication by non-contact recording and replication was required.

What problems were encountered?The copying Caligula project was one of the first we hadundertaken that involved the machining of a full bust intomarble. Until this time we had only produced reliefs by thisprocess. Initially, the replica was to be machined by auniversity spin-off company. Prior to machining starting, thecompany folded and was unable to fulfil the contract.Finding a new sub-contractor was a major task. In addition,the new subcontractor had to develop new expertise to usetheir equipment to machine into marble. Thesecomplications created a delay in the delivery of the bust;however, colour reconstruction was completed in time forthe replica to be installed in the exhibition before opening.

What were the final ‘deliverables’?A full-scale marble replica of the bust was supplied to theNY Carlsberg Glyptotek, Copenhagen. The replica requiredtwelve hours of hand finishing by our sculpture conservators.A point chisel and a flat bladed chisel were used to sharpenfacets in the hair, a drill was used to deepen the mouth, anda dremmel (small drill) was used to deepen the ears. A fineabrasive paper was employed to remove tool markings frommachining from the nose and face. To help the sculptureconservator during this process, a thin watercolour wash wasapplied to the surface of the replica, as it is difficult to seethe details on a new ‘clean white’ marble sculpture clearly.

Caligula original (right), and marble replica (left) beforepigmentation added.

Colour reconstruction on the marble replica was undertakenby the Doerner Institute, and the Glyptotek, Munich. Thereconstructed replica and the original were displayed side-by-side in Munich, Rome and Copenhagen as a part of theexhibition, ‘ClassiColor’, examining colour in Greek andRoman Classical sculpture.

All rights reserved.This case study is published with kind permission of the contributor,Vinzenz Brinkmann, Ulrike Koch-Brinkmann and Sylvia Kellner from the Doerner Institute,Dr Jan Stubbe Østergaard and Rebecca Hast from the Ny Carlsberg Glyptotek.

25

of a robotic milling machining. From our scan data, a full-scalereplica in marble was produced using 5-axis computernumerically controlled (CNC) machining. The re-pigmentedreplica was displayed, next to the original, in exhibitions inMunich, Copenhagen, and Rome during 2004 and 2005.

Instruments and softwareData capture took place in our studio (Liverpool, UK) using aModelmaker H laser scanning system. The sensor has a 40mmstripe width and was mounted on a 6-axes Faro silver arm.Sensor-object separation was maintained at 100mmthroughout. Arm calibration and sensor calibration were withinspecification. One scanning station was required and there wasno sampling during data capture. The software used to collectthe data was 3D Scanners UK Modelmaker V2 beta release.During data capture, two million points were captured. Oncescanning was complete, the data was meshed using MM V2.Rapidform2002 (INUS Technologies, Inc.) was used for post-processing and preparation of the data for machining.

Caligula: screenshots of 3D data (left – raw data, right – completed model).

The replica during CNC machining.

The final model comprised 2.3 million polygons. The rawdata is stored as CTA, SAB2 (MM file formats) and ASCIIfiles. The completed model is stored in STL format.Production of 3D fly-throughs in AVI format was undertakenin 3D Studio Max (AutoDesk Media and Entertainment).Photographic documentation was captured using a MinoltaDimage 5 3.3 megapixel digital camera at a resolution of1600 x 1200 pixels, mounted on a tripod. Each image wasmanually white-balanced. A new block of Carrara marblewas sourced from Italy. A five axes CNC machine was usedto mill the replica into a new block of Carrara marble.

Why was scanning selected?Taking a mould from the original sculpture and casting areplica could not be considered because of the delicate nature

IntroductionIn the churchyard of Prestbury Church stands an importantAnglo-Saxon Cross, thought to mark the arrival ofChristianity in the North West of England. The sandstonecross measures 940mm x 400mm x 240mm, the surface isweathered and some green moss obscures the upper east face.

The original location of the cross is unknown; however, itwas previously sitedinside the church. Thecross is highlydecorated with intricatepatterns. There areclearly sections missing,particularly between the three sections it isnow in, discernable bygaps in the patterneither side of fillsholding the piecetogether. PrestburyParochial ChurchCouncil wanted thecross accuratelyrecorded prior toconservation work andpossible re-siting.

Anglo-Saxon cross, Prestbury.

Instruments and softwareA Minolta VI 900 laser scanning system was used for datacapture. The instrument was mounted on a tripod, and set tofine mode. A middle lens was used throughout data capture.Sensor-object separation was maintained at approximately1000mm. The exception to this scanning offset was the verytop of the cross, which had to be recorded from a distance of2000mm. The calibration of the system was checked using a100mm calibration board, prior to scanning and again oncompletion of data capture. The scanner was working towithin the manufacturer’s specification. A tent was erectedover the scanning area to reduce the ambient light levels.This ensured we obtained the best data possible.

Scanning took a total of 6.5 hours and we collected 121frames. The frames were saved directly as meshes to a flash

card connected to thescanner. We undertookrough registration on- site toensure that we had coveredthe whole surface, and toensure that the datarecorded was of a highquality. Rapidform2004 PP2(INUS Technologies, Inc)was used to register andmerge the individual framesinto a coherent model. Theaverage shell-shell deviation

for this process was 0.3mm. Large areas of overlapping datawere deleted prior to merging, with the best data beingchosen wherever possible. Rapidform 2004 SP2 wasemployed for the post-processing of the data, which entailedcleaning polygons and filling small holes manually. A smallamount of localised smoothing was required in areas wherethe data was noisy. This was in the most part where there arevery dense moss patches, c 2cm2 in area. The data wasdecimated by 50% on bringing the individual frames intoRapidform 2004 SP2 prior to registering and merging, andthe final model was decimated again by 50% at the end of thepost-processing procedure. The final model containedapproximately 3.5 million polygons. Production of 3Dflythroughs in AVI format were undertaken in 3D StudioMax (AutoDesk Media and Entertainment). Photographicdocumentation was captured using a Minolta Dimage 5 3.3megapixel digital camera at a resolution of 1600 x 1200pixels, mounted on a tripod. Video footage of the cross in itscurrent location (including the immediate surroundings andthe scanning process) was recorded using a Sony 3CCDDVCAM.

Why was scanning selected?A highly accurate record of the surface of the cross wasrequired prior to possible dismantlement, conservation andre-siting. Taking a mould of the object was not an option inthis case due to the friable nature of the sandstone surface.An accurate 3D model was also required in case it is decidedat a future date to create a highly accurate replica of thecross. Laser scanning is an ideal technique for this.

What problems were encountered?The presence of moss in some localised areas meant that wecould not record the stone surface underneath as accurately asmost of the surface. However, these areas are small, and thedata clearly shows the form of the pattern in these areas.

What were the final ‘deliverables’?All raw data was supplied in CDM format. This comprisesthe 121 frames and was provided with a data log thatprovides information about eachframe. The completed data wassupplied in STL format.Photographic documentationwas provided in JPEG and TIFFformats. In addition, screenshotsof the completed data in JPEGformat, and AVI flythoughs ofthe 3D cross both textured andun-textured were supplied inAVI format.

Prestbury Anglo-Saxon crossscreenshot of 3D data.

All rights reserved.This case study is published with kind permission of the contributor.

C A S E S T U DY 6

3D Non-contact recording of an Anglo-Saxon cross, Prestburytype: triangulationkeywords: 3D, non-contact, recording, laser scanning, digitisation, 3D record, Minolta V1900 laser scanner, digital archive,Anglo-Saxon Cross, Prestbury Church

26

Scanning the cross.

27

IntroductionFor the past 50 years, a colossal red granite statue ofRamesses II, one of the mightiest pharaohs of the 19thDynasty of Egypt, has languished in downtown Cairo’sRamesses Square. The statue has been deteriorating badlyfrom pollution generated by the traffic that clogs the threemajor thoroughfares that meet at the square, and frommainline and underground trains. Some fear that the

excessive vibrations fromthis traffic may also affectthe statue in due course.The Egyptian authoritieshave decided to move thestatue to the GrandEgyptian Museum, which isbeing planned for the GizaPlateau, and should be builtby about 2010. Survey wasrequired to provide apermanent, accurate recordof the statue and to providethe data necessary to movethe massive monolith safely.(This move has now takenplace: see http://news.bbc.co.uk/1/hi/world/middle_east/5282414.stm)

Statue of Ramesses surrounded by scaffolding.

Duncan Lees of Plowman Craven & Associates (PCA) – oneof the world’s largest geomatics companies to specialise in3D survey and heritage projects – undertook the work,organised by Lon Addison of the University of California atBerkeley and UNESCO, with colleagues Björn VanGenechten from the Catholic University of Leuven inBelgium and Dr Tariq Al Murri.

Instruments and softwareThe survey used a Leica Geosystems HDS2500 laserscanner to create an accurate 3D computer model ofRamesses’ effigy. They used the scanner together with a

range of other image-based3D data collectiontechniques, includingphotogrammetry. Leica’sCyclone software was usedfor pre-processing andregistration of the scanningdatasets. The registeredpoint cloud was exportedas an ASCII .XYZ file andmeshed in RaindropGeomagic. Hole filling andchecking of the data wasundertaken in Geomagicand Cyberware Cyslice.

Registered point cloud of upper part of statue.

The mesh data was exported in .STL, .OBJ and .PLYformats. Rhinoceros software was used to produce automaticvertical and horizontal cross-sections through the statue anda contour map of the surface. Leica CloudWorx andMicrostation were used to add detail to linework elevationsof the hieroglyphs that adorn the statue.

Why was scanning selected?Traditional survey techniques such as digital data captureusing a total station theodolite or photogrammetry rely uponthe identification of edges by the surveyor orphotogrammetric plotter. The statue of Ramesses II, as withmany other statues andstructures, is an organic,irregular shape characterised bythe presence of many surfacesand few hard edges or vertices.Laser scanning is fundamentallya surface data collectiontechnique and so was ideallysuited to recording the intricatestructure of the statue in atimely and cost-effective manner.

Meshed data set depicting intricate detailing of statue.

What problems were encountered?Temperatures during the working day exceeded 45°centigrade, although the field team seemed to suffer muchmore than the equipment. The statue was surrounded byscaffolding, which proved a far from ideal base for thescanner. The HDS2500 utilised was mainly situated on thescaffolding walkways, rather than on a tripod, to minimisethe movement of the scanner.

What were the final ‘deliverables’?The mesh model was delivered to the client in a number offile formats. 3D CAD drawings in AutoCAD, andMicrostation of the four elevations of the statue, with thehard detail outlined and contourscreated, were also produced. The surveyresulted in a full record of the statue inits minutest detail, including all of thejoints, visible fault lines and cracks. Themesh will be used for a structural analysisof the component pieces of the statuebefore it is dismantled and moved,allowing calculations of the weight andvolume of the statue to be made. This inturn will supply the necessaryinformation needed to create a purpose-built secure cradle to hold the mightystatue. As it was a commercial contractthe data is held by the client rather thanby a heritage organisation.

Completed 3D model of statue.All rights reserved.This case study is published with kind permission of the contributor.

C A S E S T U DY 7

A moving story - The Ramesses II scanning projecttype: medium range time of flightkeywords: statue, Cairo, structural analysis

27

IntroductionPrehistoric rock art comprises abstract ‘cup and ring’ marksfound across many regions of northern Britain. Often, therock surface also appears to form part of the overall design.Traditionally, rock art has been recorded using 2Dtechniques, particularly photographs and rubbings. Althoughadequate for basic documentation, both techniques arelimited in terms of the level of detail and objectivity that canbe achieved, and, in the case of repeated rubbings, can beharmful to the rock surface. In addition, the limitations ofthe techniques can mislead interpretation.

The Northumberland and Durham Rock Art Project,funded by English Heritage between 2004 and 2006, isdeveloping a toolkit to enable non-intrusive digital recordingof the rock art and rock surface. The project has recruitedand trained about 50 volunteers from the local communityand the methodology has been specially designed for them to use with ease. The core aim of the project is to use thistoolkit to document all engraved panels (currently about1500) in this region and to produce a comprehensivedatabase, accessible to the general public via a website.For this ‘baseline’ recording, 2D data are captured usingGPS, digital photographs and specific recording proformas.3D data are captured using stereo photography. High-resolution recordings of select panels have also been madeusing laser scanning.

Instruments and softwareAll volunteers are using Nikon Coolpix E5400 digitalcameras. The photogrammetry methodology has beenspecifically designed for the project’s baseline recording byPaul Bryan of English Heritage, with assistance on cameracalibration provided by Dr Jim Chandler of LoughboroughUniversity. Working at a focus range of 1.5m, the level ofdetail that can be recorded using this approach isapproximately 2–3mm. The results are principally beingprocessed through Pl-3000 ‘Image Surveying Station’software produced by Topcon. It is hoped that some of this

processing will eventually be done by the volunteersthemselves, following training in December 2005. Laserscanning of five selected panels was performed byArchaeoptics Ltd using a Minolta VI 900 scanner andprocessed through Demon software. The level of detailselected for this project was 0.5mm (although the systemoffers a resolution of 0.17mm).

Why was scanning was selected?The photogrammetric technique developed for this project isuser friendly, cost effective and time efficient. For amonument type such as this, where the carved stones arerelatively small, prolifically scattered and often physicalisolated, these are crucial issues. Accessibility is alsoimportant, and the equipment required can easily be carriedby one or two people over considerable distances. Theseattributes make the technique highly suited to the volunteer-led baseline recording part of the project. In contrast, laserscanning is relatively expensive and requires specialistequipment and expertise both in fieldwork and in theprocessing stages. In this instance, it is ideally suited forhigh-resolution recording of a limited selection of panels.

What problems were encountered?The stereo photography is still in a trial phase, but the mainissues noted so far have been, first, to engage volunteers ofdiverse ages and abilities with the unfamiliar methodology.There has been a mixed response initially, with immediatetake-up by some and considerable reticence by others. Webelieve that this issue will be surmountable through a cycleof repeated trial and feedback.

The second issue is the weather and light conditions, Evenlylit images, using natural light, are preferred to the moretraditional ‘raking-light’ approach, to improve processingresults. Rain has also disrupted several recording sessions.

In terms of processing, the principal problems encountered todate relate to calibration and resolution of the chosen Coolpix

C A S E S T U DY 8

Recording prehistoric rock art by photogrammetry and laser scanningtype: photogrammetry, triangulation laser scanningkeywords: low cost, photogrammetry, Minolta laser scanner

28

Topcon’s PI-3000 software was used to create digital models. A laser scan of Ketley Crag Rock Shelter.

5400 cameras.To enable 3D measurements to be made,accurate to 2–3mm, a precise focal length is required, along withdistortion information for the camera lens. So as not to over-complicate the site work for the volunteers it was decided tocalibrate at only one focus setting – 1.5m. For rock panels ofapproximately 1m x 1m this is a suitable compromise betweencoverage and detail, but for smaller panels a shorter focus settingwould be useful to allow closer-in stereo-photography. Also the5400 camera uses a CCD with an effective 5.1 megapixels.Thisresolution is more than adequate for general usage but can limitthe processing of areas where the carved detail is less perceptiblein even, natural light. How to satisfactorily record larger panels,greater than 1m x 1m, without resorting to fixed survey targets,is still an issue for the project.

What were the final ‘deliverables’?The initial results of the stereo photography have been mostpromising. Supplied as orthophotographs (JPEG) and

surface models (currently .DXF as no .OBJ output isprovided by PI-3000) they provide a detailed, objectiverecord of the engravings and surface topography of the hostrock. These can be viewed and manipulated at will, allowingexamination of, for example, the degree and nature of lichencover, the relationship between the carvings and erosionpatterns of the rock surface, and the relative depth of theengravings. While some detail is omitted owing to the level ofresolution, the information captured using this technique hasconsiderable potential for enhancing our understanding ofthe rock art and evaluating its condition.

Finally, the exciting opportunities for public presentationrepresented by 3D data are most welcome in this contextwhere we have, on the one hand, low public awareness, andon the other, very visual but often inaccessible monuments.

All rights reserved.This case study is published with kind permission of the contributor.

29

C A S E S T U DY 9

Leasowe Man’s Skull: 3D laser scanning, digital reconstruction andnon-contact replicationtype: 3D laser scanning, digital reconstruction and non-contact replication.keywords: non-contact recording and replication, 3D laser scanning, digital reconstruction, selective laser sintering,Leasowe Man, facial reconstruction, skull

IntroductionIn 1863 workmen near Leasowe Castle in Wirral,Merseyside, found a skeleton. The skeleton has beenscientifically dated, and the remains are the only knownRomano-British skeleton from Merseyside. The remains arenow in the collections of the Natural History Museum,London, but during the summer–winter of 2005 were ondisplay in the ‘Living with the Romans’ exhibition at theMuseum of Liverpool Life (NML). As a part of theexhibition, a full facial reconstruction from the skull wasundertaken. The skull is in three parts: the cranium, and theupper and lower jaws.

The skull was scanned using a 3D laser scanner and theresulting data post-processed by Conservation Technologies,NML. The completed files were sent to the University ofManchester’s Unit of Art in Medicine for digital

reconstruction.* During this process, missing sections in thenasal and eye-socket areas were re-built. The bones of theupper jaw had previously been wired together in the wrongposition, and this was corrected. During digitalreconstruction, the upper jaw and the cranium were merged,leaving two pieces: the skull and its lower jaw. The data wasthen returned to us at Conservation Technologies, where weprepared the reconstructed data for use in replicaproduction.

Two replicas of the digitally reconstructed pieces wereproduced by selective laser sintering. One replica went to DrCaroline Wilkinson, a facial anthropologist at the Universityof Manchester, to be used in the creation of a full facialreconstruction, including skin tissue and hair. The secondreplica of the reconstructed bones was placed in the ‘Livingwith the Romans’ exhibition, where it is displayed with theoriginal skull (as a part if its skeleton). The full facialreconstruction joined the exhibition in early September 2005.

Instruments and softwareScanning was carried out in our studios in Liverpool using aModelMaker X laser scanning system with a 35mm stripewidth, mounted on a 7-axes Faro gold arm. Sensor-objectseparation was maintained at 50mm throughout datacapture. Sensor and arm calibration had an average RMSerror of 0.032mm. We required three scanning stations forthe cranium, and two each for the upper and lower jaws.

There was no sampling during data capture. The softwareused to collect the data was 3D Scanners UK ModelMakerV7 beta release. Once scanning was complete, the data were

Scanning of the lower jaw fragment with ModelMaker X laserscanning.

29

30

2D sampled at u = 0.2 and v = 0.2mm using MM V7.Polyworks V8 (Innovmetric Software, Inc) was employed fordata alignment, merging and post-processing. The maximumedge length parameter used during meshing was 1mm.Rapidform2004 PP2 (INUS Technologies, Inc) was used forregistering and merging the data from the different scanningstations together. The average maximum deviation betweenthe data from each of the scanning stations was 0.05mm.Abnormal faces were deleted and all holes were filledmanually using Rapidform2004 PP2. A small amount oflocalised smoothing was required using the paint tool.Production of 3D fly-throughs in AVI format was undertakenin 3D Studio Max (AutoDesk Media and Entertainment).Photographic documentation was captured using a Minolta

Dimage 5 3.3 megapixel digital cameraat a resolution of 1600 x 1200

pixels, mounted on a tripod. Eachimage was manually white-balanced. Digital reconstructionof the skull was undertaken usingPhantom hardware and Freeform

software (SensAble TechnologiesInc) at the University of Manchester.*

3D model of the digitally reconstructed skull.

Why was scanning selected?The remains are exceedingly friable. There was no way a castcould be taken of the skull to be used for the facialreconstruction of Leasowe man.

What problems were encountered?The digitally reconstructed data files were exceedingly large,contained much excess data, and had lost some texture onthe surface of the objects. This was a function of the software

only, and was in no way related to the digital reconstruction.To maintain as much textural detail as possible, and to createa smaller file, we combined the reconstructed and originalfiles (removing the worst areas) to obtain the best dataaround the reconstructed areas, which were not affected bythis problem, as they were created in the software rather thanimported into it. Additionally, some areas of the craniumwere too thin for the laser sintering process (replication). Thereplica would have been so fragile it would have beenextremely difficult to handle. Therefore, we added an offsetto the cranium of 3mm.

What were the final ‘deliverables’?A replica of the digitally reconstructed skull was supplied to the University of Manchester, for use in the creation of a full facial reconstruction of Leasowe man.* Another replica of the digitally reconstructed skull is on display in the ‘Living with the Romans’ exhibition at Museum ofLiverpool Life, NML.

Animated ‘fly-throughs’ of the digitally reconstructed skullwere created by Conservation Technologies and have beenused to generate public interest in the remains and theexhibition on the National Museums Liverpool website. Rawscan data in SAB2 (3D scanners file) format, raw andcompleted mesh files in STL format, photographicdocumentation in TIFF and JPEG format, the digitalreconstructions in CLY (Freeform file) and STL file format,animated fly-throughs in AVI, and project metadata, arestored at National Museums Liverpool, in line with currentdata storage guidelines.

All rights reserved.This case study is published with kind permission of the contributor,National Museums Liverpool, and Dr Caroline Wilkinson (*) from the Unit of Art andMedicine at the University of Manchester.

C A S E S T U DY 1 0

Going underground: surveying a Grade 1 listed grottotype: time-of-flight/phase comparison laser scanningkeywords: cave survey, restoration project

IntroductionThe grotto is situated within the grounds of Ascot Place inBerkshire. As you look at the rock formation from across thelake it is difficult to believe that it is a man-made structureand that it is more than 200 years old. Once you enter thegrotto, built c 1770, you see why it is now a Grade 1 listed

building. The grotto has a series of tunnels leading intodomed caverns. The walls are lined with flints and foundryslag and, in the caverns themselves, false wooden stalactitescovered in lime plaster and gypsum hang in geometricpatterns from the ceiling. Although the grotto is complete itis in need of extensive restoration and remedial works tostabilise the structure. HGP Conservation had no existingdrawings or records to work from and a detailed survey wasessential before any works could commence.

Instruments and softwareA basic topographic survey of the main outline of thestructure was carried out while creating a looped traversearound, over and through the inside of the grotto, to providecontrol stations from which the scanner targets could beobserved. Prior to surveying, it was necessary to get theexterior of the grotto cleared of the majority of the weedsView of the grotto from across the lake.

31

and light foliage covering the structure and the laurel busheswere trimmed at the bottom to give 0.5m of clear space forscanning of the ground surface.

The exterior of the grotto from the rear appears as aseries of stone steps spiralling up the grassy mound. Thefront of the grotto is a series of rocky cave entrances about ametre back from the lake edge. To be able to scan theexterior it was necessary to scan from a large number oflocations, from both short-range and from across the lake ata distance of 50–60m. The exterior was scanned with aLeica-Geosystems HDS2500 laser scanner. Approximately120 scans were collected to provide coverage over the wholeof the exterior of the grotto and adjoining waterfall. Theinterior was a different story, as the size of the chambers andthe passages meant that the scanning distance was between0.5–5m. The smallest brick chamber is in fact only 1.5m indiameter. Therefore, we decided to use the Z+F Imagerscanner to complete the interior, as its characteristics arebest suited to these short ranges.

Scan cloud of the grotto exterior.

The survey of the exterior was processed using standardsurvey methods and augmented with detail extracted fromthe external scan data using Cloudworx for AutoCAD tocreate a detailed topographic survey of the exterior of the grotto. Internally, using Cloudworx, a horizontal section was cut through the scan data at a height of c 1mabove ground, to give a wall line. All detail, including thefloor patterns, water features and seats below this point were drawn to create a floor plan of the grotto. Main ceiling features were added, including all openings, archesand stalactites.

Why was scanning selected?Laser scanning was the ideal technique to survey thisextraordinary organic structure. It has provided a unique wayfor the people that need to work with the data to work withthe survey directly and also to obtain a true three-dimensional understanding of its construction. Therenovation of the grotto will take a long time and the scandata will be invaluable throughout this process as well asproviding a permanent archive.

What problems were encountered?The size of the chambers and passages meant that theinternal scanning distance was between 0.5–5m, yetexternally the distance was up to 60m, as on the lake side it

was necessary to scan from across the lake. It was alsoimpossible to scan into every nook and cranny but thecoverage achieved was more than ample for the task.

The cramped working conditions in the interior of the grotto.

What were the final ‘deliverables’?The client was provided with a provisional copy of the floorplans so that a number of vertical sections through thestructure could be chosen. This was the first time thatanyone had an accurate plan showing the layout of thegrotto. Five primary sections were then drawn through thegrotto internally and externally. Using the survey and withcareful analysis on site, it has been possible for the exactlayout of the brick structure hidden behind its covering to befully determined. The architects and engineers arecontinuing to analyse the cloud data themselves usingPointools View and Cloudworx software. Now the restorationprogramme is beginning, additional work on scanning theadjoining cascade is being carried out and ‘stone by stone’elevations of the exterior and cascade are being drawn fromthe scan data.

Floor plan of grotto.

All rights reserved.This case study is published with kind permission of the contributor.Thanks are given to Michael Underwood of HGP Conservation for his help with this article.Images created using Pointools View.

IntroductionThe Lion Salt Works in Cheshire is the last surviving openpan salt works in the country and was recently upgradedfrom a Grade II listed building to a Scheduled AncientMonument. The building complex dates from the 19thcentury and includes office/exhibition buildings, a pumphouse, five pan houses, a smithy and a salt store. Naturally-occurring brine was originally pumped up from a depth of40m into a brine tank, which fed the evaporating pans withinthe pan houses by gravity. As the brine evaporated theforming salt crystals were skimmed off into moulds andtransported through to the brick hothouse where the blockswere removed from the moulds and left to dry in the hot airfrom the furnaces. Salt was then shipped out to Manchesteror Liverpool and exported worldwide.

The Lion Salt Works finally closed in 1986 but the trust,set up in 1993,hopes that it can berestored to a workingindustrial museumwhere brine willonce again beevaporated to makewhite salt crystals.

View of the site from the canal.

Instruments and softwareThe requirement for survey was to record the entire panhouse complex and the separate salt store, providing fullinternal and external plans, sections, elevations plus a ‘fly-through’ movie of the external point cloud. The survey isintended not only as a historical record, but also to form thebasis for a full repair specification to be prepared. Thesurvey, funded primarily by English Heritage, was carriedout as a joint venture by AEDAS (the project coordinators)and survey firm APR Services.The scanning was carried outusing a Leica-Geosystems HDS2500 laser scanner and tooka total of nine days on site, completed over three visits. Anadditional day of scanning was required towards the end ofthe project when the final ‘post plot’ was carried out. Thisfilled in small gaps in the scan data that were still toodifficult to complete by any other method. A Z+F scannerwas also trailed for a future project, the data eventually beingused to complete the interior of one of the barns.

Why was scanning selected?The project was originally specified for a photogrammetricsurvey, but a proposal to use laser scanning to carry out the

majority of thesurvey was acceptedby the Trust andEnglish Heritage asa suitable way torecord such a seriesof structures, whichhave virtually nostraight lines.

What problems were encountered?A point density of 10mm on the inside and 10–15mm on theoutside of the buildings provided sufficient coverage to drawthe individual timbers required. However, this generated avast amount of data (several gigabytes) which, for ease ofhandling, was divided up into the internal point clouds foreach pan house and a single point cloud for the entireexterior. Scanning was not always easy to complete, as insome areas buildings had collapsed.

It was necessary to use cloud-to-cloud registration toregister a few of the scans for areas that were either unsafe orinaccessible to place targets. This software process joinsoverlapping scans together by comparing and aligning thesame geometry within the overlap. This registration processwas carried out while on site to ensure coverage, and thatsufficient control had been observed.

Interior of salt barn.

What were the final ‘deliverables’?More than 40 elevations, plans and sections were producedusing either Cloudworx for AutoCAD or APR Services’Pointools software. Both programs enable the user to viewand manipulate the point cloud to produce either 2D or 3Ddrawings. In order to create the floor plan, a datum heightwas chosen for each barn. A horizontal section was cutthrough the point cloud to create a plan at that height. Agrid of levels over each floor was extracted from the pointcloud along with all beams and main roof timbers.

Elevation to pan houses.

Internal and external elevations were drawn by isolating theindividual wall being drawn from the main point cloud andtracing the detail, cutting sections through the cloud wherenecessary to verify the structure being drawn. Although this isnot a fast process, it is probably the quickest method presentlyavailable to convert point cloud data to 2D vector drawings.All rights reserved.This case study is published with kind permission of the contributor.

C A S E S T U DY 11

Surveying industrial archaeology: scanning Lion Salt Works, Marstontype: time-of-flight/phase comparison laser scanningkeywords: industrial archaeology, restoration project

32

The interior of a salt barn.

33

C A S E S T U DY 1 2

Scanning at the edge: assessing the threat of coastal recession atWhitby Abbey, North Yorkshiretype: Airborne Laser Scanning, terrestrial laser scanning (time of flight)keywords: geohazards, monitoring, management plans

IntroductionFounded in 657 AD, Whitby Abbey occupies a prominentheadland site, overlooking the historic town of Whitby. Thepresent remains date back to the 11th Century and aresituated alongside associated buildings of historical andreligious importance, including St Mary’s Church(established in the 12th Century). The site is cared for byEnglish Heritage, and an engaging visitor centre helps todraw large numbers of tourists to the site every year.

The site is under threat from coastal recession.

However, part of the reason why this site is so attractive, isalso the root of its vulnerability. Highly visible from bothland and sea for miles around, the very existence of thisancient religious site is threatened by the ever-encroachingcoastline. The headland is characterised by near-vertical cliffs c 60m high. While the abbey itself is still some 160mfrom the cliff edge, parts of the site have started to suffer the effects of cliff erosion. In 2000 a significant cliff collapseoccurred, prompting English Heritage’s (then) Centre forArchaeology to excavate and document importantarchaeological deposits near the cliff edge. Although this may have alleviated immediate concerns, coastal erosion isan incessant natural process, and it is important thatorganisations such as English Heritage can gain an improvedunderstanding of the longer-term threat so that, if necessary,steps can be taken to preserve this valuable heritage.

Instruments and softwareAirborne Laser Scanning (ALS) was flown for the headlandarea in April 2005 and again in August of the same year. Thismonitoring strategy was designed to examine the capabilitiesof ALS for detection of change due to coastal erosion.

The ALS was captured by NERC’s Airborne Researchand Survey Facility (ARSF) using an Optech ALTM 3033scanner. This instrument has a specified absolute accuracy of± 150mm, and is capable of delivering first and last pulsereturn data. The site was flown at an altitude of 1000m,resulting in a swath width of c 680m. This produced data at a

resolution of approximately one point per square metre, andthe headland area and inter-tidal zone were comfortablycaptured in three overlapping flight-lines.

Data was collected at one point per metre square.

The raw laser ranges were processed by the ARSF toproduce XYZ point files each containing several millionpoints. Further processing of this data will be performedusing TerraSolid’s TerraScan and TerraModeler software.

Why was scanning selected?ALS enables continuous coverage over large areas, and is an excellent tool for the rapid acquisition of digital elevation data. It does not suffer from some of the problemsassociated with photogrammetry, such as the lack of detail in shadow areas, or correlation problems due to poorimage texture in areas such as beaches and foreshores.In addition, although ALS is restricted to predominantly dry conditions (to prevent laser returns from raindrops),it is much less weather reliant than image-dependanttechniques such as aerial photography. With on-boardpositioning and attitude sensors for geo-referencing, ALSdoes not generally require much ground effort in terms ofcontrol points, other than one or more GPS base stations (in this case data from the Ordnance Survey’s ActiveReference Station network was provided).

What problems were encountered?Problems associated with ALS surveys are often connectedto poor reconnaissance and preparation. The number andspacing of flight lines, flying height and desired spatialresolution of the data are all essential considerations that hadto be balanced. Gaps between adjacent flight lines can be aproblem if the correct overlaps are not applied. Satellitevisibility is another concern, and appropriate GPS missionplanning is essential.

Although the aerial dataset is capable of deliveringcomprehensive, continuous coverage at a fairly highresolution, it tends to suffer from occlusion problems inareas of steep slopes and overhangs, such as those found onthe lower half of the cliffs at Whitby. In spring 2006

33

34

terrestrial laser scanning will be used to fill in such datagaps, and provide high resolution modelling of the complexcliff surface.

The April 2005 ALS survey comprised 4.5 millionpoints. This data volume presents significant challenges, notonly to data processing, but perhaps more importantly todata storage and archiving.

What were the final ‘deliverables’?Digital Terrain Models (DTM) will be produced by filteringout vegetation, buildings and other artefacts. The DTMs canthen be used for comparison and analysis of multi-temporaldatasets in order to detect areas of change, which may be aresult of coastal recession.

The data sets will be fused together to allow comparisonsbetween the different temporal datasets. This will allow theproduction of maps highlighting the magnitude and locationof changes to the terrain over time. Analysis of thisinformation alongside historical records, such as maps andaerial photographs should enable assessment of the rate of

cliff recession at Whitby Abbey Headland. This should proveuseful to English Heritage in planning of future preservationworks for the site. Work will continue on this project over thecoming year, with final results expected in late 2006.

Digital terrain models will be produced.

All rights reserved.This case study is published with kind permission of the contributor.

C A S E S T U DY 1 3

Automatic digital reconstruction of a roof truss:laser scanning and automatic extraction of a roof truss in the St. Petri Cathedral Bautzen, Germanytype: time of flight scanning, Development of point-cloud algorithmskeywords: terrestrial laser scanning, roof trusses, automatic extraction, structural analysis

IntroductionUnder the high saddle roof of the St. Petri cathedral a trussis located which consists of five floors. Terrestrial laserscanning was used to record the geometry of the bottomfloor, which had dimensions of 60m x 31m x 5m.

St. Petri Cathedral, Bautzen, Germany.

The survey aimed to collect data for an automatic digitalreconstruction and structural analysis of the truss itself.Although the point cloud contains a huge amount ofgeometrical information, many users prefer a 2Drepresentation for interpretation. Therefore, the survey also aimed to derive a general 2D plan from the point cloud. Algorithms were developed for an automatic analysisof the scanner data.

Instruments and softwareA Riegl LMS-Z420i laser scanner was usedfor the project. The scanning process iscontrolled by a notebook and Riegl’sRiSCAN PRO software. This panoramic laserscanner has a field of view of up to 80° x

360°. The range finder, based on the principleof pulsed time-of-flight, is able to recorddistances between 2m to 800m with anaverage accuracy of ± 7.5mm.

Terrestrial Laser-scanner Riegl Z420i.

Scanning configurationBecause of numerous beams located inside the truss thereare many occlusions. The resulting scan shadows werereduced by increasing the number of scanning positions. Theselected configuration was a network of 12 scans. At each ofthe 12 positions a panoramic scan with an angular resolution

35

of 0.1° was collected. Thus, about three million points weremeasured at each laser scanner position, which results in 35million points for the first floor.

Each individual scan had to be transferred into a uniformproject co-ordinate system. For this, circular and cylindricalretro reflectors were distributed in the observation area, toserve as tie points. More than three of these tie points wererequired in adjacent point clouds to obtain a transformation.

Why was scanning selected?Laser scanning provides the best potential for automation ofthe recording and analysis of the measurement data. It is alsoless time-consuming than conventional techniques such astacheometry or manual measurements.

Furthermore, old buildings often have unstable floors,which result in vibrations that limit the accuracy of themeasurements. For example, the truss of the St. PetriCathedral consists of timber floorboards. Working with atachymeter was not practicable because of the operatormoving around the tripod. Using the laser scanner withwireless data transfer solved this problem.

What were the final ‘deliverables’?At first the point cloud was cut and projected onto horizontal2D layers. An algorithm to segment and model the objects(which are mainly rectangles) was also developed. This wasbased on the identification of lines using a Houghtransformation. The result of this method was a 2D plan of thefirst floor. Furthermore it is possible to produce maps indifferent height levels to extend this approach from 2D to 3D.

A second method, developed as part of the project,segments and models the 3D point cloud directly using aHough transformation which identifies planes within thepoint cloud. The intersection of these planes results in 3Dmodels of the objects.

Approximately 60% of the roof truss could be modelledautomatically. As a by-product of these calculations it waspossible to extract the topology of the whole truss which canbe used together with the geometry for a structural analysisof the truss static.

Further information from: Henze, Wulf-Rheidt, Bienert,Schneider, D: Photogrammetric and geodetic documentationmethods at St. Petri Cathedral, Bautzen, Paper presented atCIPA Symposium 2005, Turin (Italy).

All rights reserved.This case study is published with kind permission of the contributor.

360° x 80° intensity image within the roof truss.

Analysis of a intersection plane: a) intensity image;b) 2D point cloud; c) 2D modelling.

Automatically generated map of the floor plan.

IntroductionThe Witham Valley Research Project has utilised varioussurvey techniques and part of the designated area hadalready been subject to archaeological interpretation andmapping from aerial photographs. It was decided to compare LiDAR with the data recorded from the traditionalaerial survey methodology employed by LincolnshireNational Mapping Programme (NMP), with particularinterest in the usefulness of available archive data flown fornon-archaeological purposes.

Instruments and softwareLincolnshire County Council provided 2m resolution LiDARdata collected with an Optech ALTM 3033 system by theEnvironment Agency (EA) in March 2001. Based onprevious experience with satellite and multi-spectral imagerythis was not expected to yield the same level of results as that from the Stonehenge survey. The data was providedin 2km by 2km ASCII gridded files based on the OS grid,which could be used directly with ArcGIS. Because thelandscape of the Witham Valley is generally very flat,and a lot of the features had been severely reduced byploughing over several decades, it was necessary toexaggerate the height ratio in the images up to 20 times tomake features evident.

Why was scanning selected?The EA LiDAR data was chosen to test the usefulness ofstandard data not captured for archaeological purposes.

What problems were encountered?An initial problem was that the interpretation was carried out in the English Heritage York office, where the initialNMP project work had been carried out, while the tools forLiDAR manipulation were only available in Swindon. This

meant that the LiDAR data was processed in Swindon, afterwhich JPEG images of individual tiles were sent to York.

These images were examined in York and compared with known NMP data and if they were thought to have the potential for additional features that were notimmediately visible on the image provided, requests weremade to produce new images at a different azimuth orelevation etc. This was not the ideal method for carrying out the survey, but proved workable and produced someuseful results.

The second problem was that, as expected, the 2mresolution was insufficient to show any but the largestfeatures, such as field banks.

Barlings Abbey with NMP data overlaid (LiDAR courtesy of Lincolnshire County Council; source, Environment Agency,March 2001).

What were the final ‘deliverables’?The key results from this survey were not the DEMs fromthe LiDAR, but the interpreted overlays that were comparedwith previous surveys.

StixwouldNear Stixwould the LiDAR data revealed a length of bank more than 750m long, which had previously been seen as a cropmark and recorded as a possible length of Romanroad, but was later discounted as a probable field bank ordrainage feature. The LiDAR data combined detail with context to produce another possible interpretation for the feature. The detail showed that this was an extensivebroad banked feature at a dramatically different alignment to other field boundaries in the area, and so was clearly not an old field bank or drainage feature. At its eastern end the bank turned sharply to the north and headedstraight towards the site of the former Cistercian priory.There are a number of other known causeways providingaccess to the abbeys and priories in the valley and this may be yet another example.

The context given by the broader DTM shows that thisroute leads to the main valley bottom, but avoids the lowerlying and potentially wet areas.

C A S E S T U DY 1 4

Airborne LiDAR for ancient landscapes: assessing LiDAR formapping large historic landscapestype: airborne laser scanningkeywords: Witham Valley, landscapes, aerial survey, National Mapping Programme (NMP)k

36

Southrey shrunken settlement? (LiDAR courtesy of LincolnshireCounty Council; source, Environment Agency, March 2001).

Stixwould (LiDAR courtesy of Lincolnshire County Council;source, Environment Agency, March 2001).

Bardney environsA second example can be used as a cautionary tale toemphasise that LiDAR data simply reflects differences in theheight of features on the ground; it draws no distinctionbetween a prehistoric barrow and a modern electricity pylon.It is simply one more source of data and must always beused in conjunction with other available information.Examination of features in the vicinity of Bardney revealedan interesting roughly ‘playing card’ shaped feature. This waschecked against the current OS base map and the 1stEdition and was visible on neither. Indeed the 1st Editionmap even seemed to show a field boundary that has sincebeen removed as respecting the line of the feature.

Examination of the site in its landscape context(particularly with the height exaggerated ten times) revealedthat it was on a slight ridge with a commanding view overthe valley below. The combination of its size and shape andits location gave clear indications that this might be apreviously unknown Roman fortlet or signal station.However, further examination of other sources revealed adifferent story. It was noted when examining the current OSmap that the site lay on the edge of a former airfield.Inspection of aerial photographs from during andimmediately after WWII showed that this was an area of hardstanding leading to a possible hangar or storage building.

Combined LiDAR 2001 and aerial photo 1946 (LiDAR courtesyof Lincolnshire County Council; source, Environment Agency,March 2001; /RAF 3G/TUD/UK 197 Part VI 5449).

All rights reserved.This case study is published with kind permission of the contributor.Data for the project was provided by Lincolnshire County Council.

37

C A S E S T U DY 1 5

Forest of Dean: LiDAR for mapping historic landscapes in woodlandtype: airborne laser scanningkeywords: Welshbury, landscapes, aerial survey, National Mapping Programme (NMP)

IntroductionSince 2000 the Aerial Survey team at English Heritage hasbeen examining LiDAR data with a view to assessing itssuitability. A fresh aspect of LiDAR and new potential wasrecognised in 2003 when the possibility of using last pulsedata was pointed out in a presentation by SimmonsAerofilms. It was realised that the ability to penetrate treecanopies could be extremely useful in revealing features inareas where traditional aerial survey was unsuitable.

The Forest of Dean had been subject to standard aerialsurvey techniques as part of the National MappingProgramme (NMP) for Gloucestershire. Using historic aerialphotographs taken over a number of years it had beenpossible to record some features that were visible duringperiodic phases of felling. But given the nature of the Forest,with a high proportion of land covered by dense woodland,there were large areas where very few archaeological featureswere recorded.

The Aerial Survey team became involved in a projectwith the Cambridge University Unit for LandscapeModelling (ULM), Forest Research at the ForestryLea Bailey Wood, Forest of Dean NMR23322/02 (07-Nov-2003).

37

38

Commission and Gloucestershire County CouncilArchaeology Service to look at the Iron Age hillfort atWelshbury. The site had previously been recorded on theground by the Royal Commission on the HistoricalMonuments of England (RCHME), but when the area wascovered by the NMP project, very little detail wasrecoverable because of the density of the vegetation. It wastherefore seen as an excellent site on which to test thecapabilities of LiDAR.

Instruments and softwareAn airborne LiDAR survey of the site was carried out inFebruary 2004 using the ULM’s Optech ALTM 3033system. The details of their project are recorded in Devereuxet al (2005), from which the following technical specificationsare also taken. Ground GPS support was provided by a dualfrequency, Novatel receiver located at an Ordnance Surveypassive recording station. The maximum distance from thebase station to the most extreme point on the survey site was28km whilst the shortest distance was 1.2km.

Two separate surveys of the site were conducted togenerate approximate point densities of four per squaremetre and one per square metre. The size of the laserfootprint was set to a nominal 0.8m for the four points permetre survey and 1.25m for the one point per metre survey.By flying the surveys during winter, the deciduous canopywas devoid of leaf cover and the understorey vegetation wasat a minimum, thus ensuring maximum laser penetration tothe ground surface.

The survey point cloud data were converted to a 0.25mand 1m grid (for the high and low resolutions surveys,respectively) by assigning cells with the point value of thelaser observation that falls within the cell. Where more thanone laser observation was found in a cell the last oneencountered in the point cloud was used. Empty cells werefilled by smoothing their neighbours. Images were collectedfor first pulse, last pulse and intensity (the overall strength ofthe laser return). Staff at ULM wrote a vegetation-removalalgorithm to create a digital elevation model (DEM) of thetopography of the site under the forest canopy (Devereux et al 2005).

As there is no mathematical expertise within the AerialSurvey team for the writing of algorithms it was feltimportant to analyse the potential of using just raw last-pulsedata to see what information could be gained that was notavailable from the first pulse. The data was provided toEnglish Heritage in the form of ASCII tables recording the

XYZ and intensity data for the first and last pulse in a singletable. This was separated into tables that could be read intoArcGIS 8.3 where the 3D Analyst module was used tointerpolate to raster using Inverse Distance Weighting.

The results were very positive in that although they didnot remove all traces of vegetation, as was achieved by thealgorithms, they did reveal a large amount of previouslyunseen detail.

Welshbury Hillfort: LiDAR last pulse.

Why was scanning selected?The potential of LiDAR to penetrate the canopy and allowthe recording of features that were invisible to standard aerialphotographic techniques made it ideal to test in such anenvironment that is also difficult to survey on the ground.

What problems were encountered?The Forest of Dean was the first survey area where theAerial Survey team had direct access to the LiDAR data andthis led to a very steep learning curve in how best to use thedata. The large file sizes also created practical difficulties interms of the processing power of the team’s PCs.

On a more technical note, while the processed algorithmleft a bare earth DEM the raw last pulse data left a largeamount of ‘stumps’ representing either the actual trunk ofthe tree or particularly dense areas of foliage that could notbe penetrated. This was particularly noticeable in the area ofconifer plantation where even the last pulse data was unableto penetrate the canopy owing to the density of foliage. Thisdata is useful for more detailed analysis as it providesinformation to aid location for follow-up fieldwork and onthe condition of the archaeological features.

What were the final ‘deliverables’?While the ULM algorithm produced a true DTM the rawlast-pulse data produces something between a DTM andDSM as it removes the bulk of the vegetation, but not all ofit. This was illuminated from various elevations and azimuthsto reveal variations in the surface that might relate toarchaeological features. Because the LiDAR coverageextended beyond the edges of the woodland it was possibleto compare the results with those from the conventionalNMP survey and confirm the presence of known features.

Devereux, BJ, Amable, GS, Crow, P and Cliff, AD 2005,’The potential of airborne LiDAR fordetection of archaeological features under woodland canopies’ Antiquity 79, 648–60

All rights reserved.This case study is published with kind permission of the contributor.Theoriginal survey was carried out by Cambridge University Unit for Landscape Modelling andspecial thanks are due to Bernard Devereux and the staff at ULM.Welshbury Hillfort: LiDAR first pulse.

39

IntroductionThis was a collaborative project involving the NationalElectronics and Computer Technology Center (NECTEC),Bangkok, Thailand, and The Department of Spatial Sciencesat Curtin University of Technology, Perth, Australia. The aimwas to create a realistic and accurate 3D model of thisculturally and architecturally significant heritage precinct andto make it accessible over the WWW, thereby providing thearchaeological as well as heritage management community,and the general public with the opportunity to examine andvisit it remotely. Two sites were recorded: Wat Mahathat inAyuthaya in March 2003 and Wat Mahathat in Sukhothai inNovember 2003.

The project tasks were divided between the two groups.The Curtin group were responsible for:l laser scanningl control surveys for point cloud integrationl data management

While the NECTEC group undertook:l surface modellingl texture mappingl web publishing (including the required multimedia

functionality)

Instruments and softwareFor the scanning, Curtin’s Riegl LMS-Z210 was selectedsince, at the time of project’s initiation, it was among thefaster instruments offering a nearly-complete horizontal fieldof view. Rapid and complete data capture was very importantsince both sites are popular tourist destinations and it wasrequired that our activities would not be a disruption.

The I-SiTE software was used for data acquisition, pointcloud editing and registration. Geomagic Studio was used forsubsequent surface modelling and texture mapping.Software developed by the Curtin University group was usedfor network adjustment of the surveying data to control pointcloud registration.

Why was scanning selected?Scanning was selected basically for its ability to rapidlycapture very dense, accurate 3D datasets.

Large Buddha statue at Wat Mahathat, Ayutthaya.

What problems were encountered?Problems encountered in the field and during processingincluded:l providing adequate battery power for the scanner and

laptopl the design of a network to maximise data coverage and

minimise data shadows due to occlusions caused by themany structures on site, eg, chedi, prangs, etc

l planning work around the activities of the tourists on sitel during processing, handling of the merged point cloud of a

dozen or so scans in the I-SiTE software (which waslimited at the time)

What were the final ‘deliverables’?A texturemapped 3Dmodel of eachsite was the finaldeliverable.

Texture mapped3D model of theBuddha statue.

C A S E S T U DY 1 6

Generating accessible 3D models: recording UNESCO WorldHeritage Sites at Ayuthaya and Sukhothai,Thailandtype: ground based laser scanningkeywords: archaeological sites, web dissemination, textured models

The Wat Mahathat precinct within the Sukhothai site.

IntroductionThe Discovery Programme’s Medieval Rural SettlementProject is undertaking an excavation at an earthwork moundin the village of Tulsk, Co Roscommon, Ireland.

The excavation is revealing at least three distinct phasesof activity on site, including the remains of a large masonrytower. Work began in 2005 with the task of excavatingthrough the large amount of rubble that filled the tower’sinterior. The tower measures some 20m long and 10m wide,and had rounded corners and a battered external wallprofile. It seems to have been destroyed by the late 1500s, atwhich time the mound was reoccupied. The refortification ofthe mound might be attributed to the presence of SirRichard Bingham, Queen Elizabeth’s Governor in Tulsk inthe 1590s.

Instruments and softwareIn advance of excavation a digital elevation model (DEM) ofthe site was created by DGPS survey of over 20,000 heightpoints (c 1m spacing) referenced to the Irish Grid. GPSprocessing was undertaken using Trimble Geomatic Officesoftware, with the DEM created using the ESRI’s 3D Analystsoftware.

A DEM of the site in advance of excavation surveyed by DGPS.

A Mensi GS101 laser scanner, controlled by Pointscape 3.1software hosted on an Itronix pen computer was used toscan the excavation surfaces. Each surface was recorded bybetween four and six scans, depending on the size andcomplexity of the surface to be scanned, with the objectivebeing to minimize shadow areas on the scans. Scanresolutions were generally 5mm (at 10m), giving scan timesof approximately 20 minutes, generating data sets of c 200MB. A portable electrical generator was used to providea constant reliable power source for the digital equipment.

A fixed network of seven control spheres was establishedaround the excavation site, positioned to allow at least fourspheres to be seen from any scanner set up. Surroundingthis, a series of reflectorless survey targets were used to placeall scanning within a pre-known georeferenced framework.

Registration of scans was done in Realworks Survey 5.1,using the automatic registration function. Georeferencing ofthe registered scans was also done in Realworks. Orthometricviews of the RGB point cloud were generated and output as

high detail TIFF images. The resulting orthometric imageswere adjusted in Adobe Photoshop to enhance imagecontrast and brightness. The images were then converted toGeoTIFF images using GeoTIFF Examiner software, byapplying the pixel scale factors and world X and Y tie pointsas provided by Realworks as an associated text files. TheGeoTIFF images were opened in ArcView 9.1 GIS softwareand displayed with all other relevant site survey data, forexample trench grids. Resulting scaled plots from ArcViewwere printed and laminated to allow field completion andinterpretation by the site supervisor to be marked before theexcavation proceeded to the next level.

Why was scanning selected?The first two seasons of excavation used conventional planand section drawing to record the excavation, processes thatrely on pencil drawing – a time consuming, highly subjectivemethod that has a low level of accuracy and a high level oferror. The excavation was revealing large elements ofcomplex stone work, which the graphic survey, using tapesand planning frames, was not satisfactorily recording.

We believed that scanning would be able to provide amuch improved quality of record in a shorter time.Additional benefits of the scanning process would be the 3Dnature of the data, which would replace the need forconventionally surveyed spot heights, and which would aidvisualization and interpretation in the post-excavation phase.

What problems were encountered?The Mensi scanner has a limited vertical field of view sosome of the scan set ups required tilting of the instrument toview down into the trenches. Care was needed to maintainthe instruments balance, and that it could still scan the targetspheres. It was originally planned to operate through awireless connection between scanner and control software,but this proved unreliable and was replaced by a fixed 5mnetwork cable. Problems also persisted with network

C A S E S T U DY 1 7

Earthwork excavation: scanning archaeological excavationstype: terrestrial laser scanningkeywords: long range laser mapping system, time of flight measurement

40

The Mensi GS scanner set up among the masonry remains ofthe tower.

connection cables displaying irregular interruptions betweenthe scanner-to-computer connection. In the field theproblem of commonality of control spheres was solved byusing a template job, which contained a station set up withall seven spheres observed. In this way any four could beobserved from each subsequent station set up.

Weather restrictions were a problem, with the laserscanner not being useable in the rain (due to droplets on thewindow, and interference of the beam).

What were the final ‘deliverables’?The georeferenced orthometric TIFF images of plans andsections are considered our basic deliverables during the fieldphase of the excavation. These can be produced ready forfield verification within 30 minutes of the final scan beingcompleted. This rapid turn around is a vital part of theprocess as it minimizes the down time of the excavationteam, and allows the record to be completed in the fieldbefore further excavation takes place. Initial scepticism fromthe site supervisors was overcome by carrying out acomparison between the laser scanning and conventionalhand drawn planning methods on the same surface. Theimproved speed and quality were immediately obvious, andthe excavation team was excited and supportive about theimplementation of the laser scanning method.

Beyond the immediate field deliverable the challenge is tomerge the point cloud records with the finds positioned by

total station survey into a useable GIS ‘project’. While seeingthe immediate benefit of ‘flattening’ our data to produceplans and sections, it is important that we generate the addedvalue from the three-dimensional data of our record as well.Assessment of the various point cloud meshing software iscurrently underway to enable the creation of correct anddetailed surfaces.

The final deliverable result.

All rights reserved.This case study is published with kind permission of the contributor.

41

42

C A S E S T U DY 1 , 2 , 3 , 4 , 5 , 6 & 9Conservation Technologies, Conservation Centre, Whitechapel, Liverpool, L1 6HZTel: +44 (0) 151478 4916 Fax: +44 (0) 151 478 4810 conservationtechnologies@liverpoolmuseums.org.ukwww.conservationcentre.org.uk/technologies

C A S E S T U DY 7Plowman Craven & Associates141 Lower Luton Road, Harpenden, Hertfordshire, AL5 5EQwww.plowmancraven.co.uk

C A S E S T U DY 8Tertia Barnett, Northumberland and Durham County CouncilsPaul Bryan, English HeritageDr Jim Chandler, Loughborough University of Technology

Case studies contributed by:

C A S E S T U DY 1 0 , 1 1APR ServicesUnit 6a, Chaseside Works, Chelmsford Road, Southgate, London, N14 4JNTel: 020 8447 8255 Fax: 020 8882 8080mail@aprservices.net www.aprservices.net

43

C A S E S T U DY 1 2Pauline Miller, Jon Mills and Stuart Edwards, University of Newcastle upon TynePaul Bryan, English HeritageStuart Marsh and Pete Hobbs, British Geological Survey

C A S E S T U DY 1 3Technical University DresdenInstitute of Phototgrammetry and Remote Sensing, Prof. Dr. H.-G. Maas01062 Dresden, Germanydanilo.schneider@tu-dresden.de www.tu-dresden.de/ipf/photo

C A S E S T U DY 1 4 , 1 5Simon Crutchley, English Heritagesimon.crutchley@english-heritage.org.ukwww.english-heritage.org.uk/aerialsurvey

C A S E S T U DY 1 6Dr Derek Lichti and Jochen FrankeCurtin University of Technology, GPO Box U1987, Perth, WA 6845, Australiawww.cage.curtin.edu.au/lascan/index.html

C A S E S T U DY 1 7Robert Shaw, The Discovery Programmerobert@discoveryprogramme.iewww.discoveryprogramme.ie

Published July 2007

© English Heritage 2007Edited and brought to press by David M Jones, English Heritage PublishingDesigned by www.mark-simmons.comPrinted by Wyndeham Westway Ltd

Printed on recycled paper containing aminimum of 75% recovered fibre, theremainder being from sustainable sources.

Product Code 51326

English Heritage is the Government’sstatutory advisor on the historicenvironment. English Heritage providesexpert advice to the Government on allmatters relating to the historicenvironment and its conservation.

For further information and copies of thisleaflet, quoting the Product Code, pleasecontact:

English HeritageCustomer Services DepartmentPO Box 569Swindon SN2 2YP

Telephone: 0870 333 1181Fax: 01793 414926E-mail: customer@english-heritage.org.uk

Text compiled by Dr David Barber andProf Jon Mills, School of CivilEngineering and Geosciences, NewcastleUniversity

AcknowledgementsAll project associates are thanked for theirinvolvement, in particular those who havecontributed case studies and those whoattended steering committees: Paul Bryan(English Heritage), Tony Austin(Archaeology Data Service), Bill Blake(English Heritage), Wolfgang Böhler(i3Mainz), Alistair Carty (ArcheopticsLtd), Martin Cooper (National MuseumsLiverpool), Tom Cromwell (EnglishHeritage), Ian Dowman (UniversityCollege London), Peter Horne (EnglishHeritage), Graham Hunter (3D LaserMapping Ltd), Alan Leith (RoyalCommission on Ancient HistoricMonuments of Scotland), Keith May(English Heritage), Faraz Ravi (PointoolsLtd), Pete Stonehouse (Viriditas UK Ltd).

Cover figure: Laser scanning systems andtheir output datasets, applied to the full rangeof heritage subjects.