Utilization of LiDAR and IKONOS Satellite Data for Security Hotspot Analysis based on Realism of 3D City Model

Mazlan Hashim, Maged Marghany, Mohd Hafiz Anuar and Mohd Rizaludin Mahmud

Institute of Geospatial Science & Technology (INSTEG)INSPIRING CREATIVE AND INNOVATIVE MINDS

Introduction

• Geospatial applications for national security concerns– Social unrest– Terrorism motives

• Some of the important applications : – homeland security, – border monitoring, – battlefield intelligence, and – sensitive facility monitoring

Enabling technology photo-realism in 3D

• Stereo aerial photography• Reconstruction of geometry • Rendering radiometry• Visualization

– 3D and virtual reality

Enabling technology photo realism in 3D

• Stereo aerial photography– Fine spatial resolution satellite image – LiDAR

• Fused LiDAR and IKONOS

• Reconstruction of geometry • Rendering radiometry

– Real life texture

• Visualization– 3D and virtual reality– Decision making with multi-criteria for security

hotspot analysis– Surrounding effects – rain effects

Existing

Introduce

The context

Past

Now

Problem

production of photogrammetric-based 3D city and their analysis is a task/ business of “photogrammetric+3D visualizer”

Wide spread availability of statistical graphics and GIS software

Appearance of visualization and mapping services in GIS and related enabling technology

Photo realism not for to novice and casual users Thematic mapping and decision support tools become

principally available to novice and casual users, but Many potential users lack the required knowledge to

utilize the tools in new applications

Our Study

• Method of utilizing 3D city urban model to assess possible shooting location for security hotspot detection

• Materials– LiDAR (Leica ALS50, DSM gridded 25cm)– IKONOS satellite data (pan-sharpened)– In situ GPS for GCP– Meteorological data

Study Area• Merdeka

Square, Kuala Lumpur

LIDAR• Leica AL50 LiDAR scanner

system• Aircraft speed: 70m//s• Flying height: 1500 m above

ground level• Scanner field of view (half

angle): +16o

• Scan frequency: 14 Hz• Swath width: 863 m (600 m

with a 30% sidelap)• Pulse repetition rate: 10 kHz• Sampling density: average 2.4

m

FiIn situ GCP collectionTextures for rendering

Method

Systems:•Erdas Imagine V9.2•Arc GIS•Arc Scene•Google Sketup5

Data Pre-processing

Two tedious merging tasks:2)Merging Images with different sensor

geometry (spatial)– Use LiDAR as master (navigation

information)– Use spatial domain fusion (varible domain

fusion, high frequency domain modulation)

Data Pre-processing

Two tedious merging tasks: (cont’d)2) Spectral sharpening - Ensure no artifacts (visual and spectral

information ?

Simulated effective range

Desert eagle: Effective range 200m(magnum)

Dragunav: Effective range 800m

M16A: Effective range 450m

δ

Scores and rank of options

Crit 1: viewshed

0 0.5 1.0

Poor Excel

Crit 2 Eff dist

0 0.5 1.0

Poor Excel

Crit 3 weapon type

0 0.5 1.0

Poor Excel

Crit 4: Sniping Quality

0 0.5 1.0

Poor Excel

Crit 5: Rainfall effects

0 0.5 1.0

Poor Excel

Opt 1 Opt 2 ….. Opt n

Crit 1 0.33

Crit 2 0.18

Crit 3 0.88

Crit 4 0.90

Crit 5 0.15

OverallScore x

AssigningWeights tocriteria

Aggregating values

The ranking, weight and normalized weight

Distance(meter)

Direct ranking Weight

Normalized Weight

Normalized weight to 100

50 1 16 0.117647 12100 2 15 0.110294 11150 3 14 0.102941 10200 4 13 0.095588 10250 5 12 0.088235 9300 6 11 0.080882 8350 7 10 0.073529 7400 8 9 0.066176 7450 9 8 0.058824 6500 10 7 0.051471 5550 11 6 0.044118 4600 12 5 0.036765 4650 13 4 0.029412 3700 14 3 0.022059 2750 15 2 0.014706 1800 16 1 0.007353 1

136 1 100

Viewshed

N

Az (0 ~360) Horizon, 0o

90oTarget

Viewing angle

Viewangle

Dirrank

Wt

110

110

max

:::::::::::::

4546

400

Min0

::::::::::

0

90 0 0

Multi-criteria for ranking risk

W = n – rj +1Σ (n- rk +1)

where:W is the normalized weight for the jth criterion;n are numbers of criteria under consideration; andrj and rk are rank position of the criterion. ( k = 1,2,…,n).

Implementation ease, and simplicity of reasonings of output to decision-makers:

Rank viewshed + effective distance for the 3 weapon types

Producing 3 risk maps

Refined for sniping quality ratio

Simulated rainfall effects

Sniping quality ratio

N

AzHorizon, 0o

90oTarget

Viewing angle

P

A

B

Effective view, elliminating blind spots, etc = angle APB=180o;

Account both viewing angles (locus!)

Ratio

(degree/180) Weight

below 0.25 1

0.25 < x <0.5 2

0.5 < x <0.75 3

0.75 < x < 1 4

Rainfall effects

Target heavy medium drizzle

viewer

Rain amount(mm)

VisibilityDistance (m)

Dir ranking

No rain (clear day)

800 1 (max)

Heavy 150 4 (min)

Medium 300 2

Light (drizzle)

500 3

ResultsCity block

City block with arbitrary textures

Textures from real photos can be draped on city block model

ResultsConstructed city blockwith some real draped texture Top view

Constructed block, side view (east)

Complete texture compiled,draped on 3D block

Results

viewshed analysis draped on the Lidar data

Visible sitting areas from vantage points (white building tops)

Snipping effective range from centre of square

Desert eagle: 200m M16A: 450m Draganov: 800m

Location of potential risk security hotspots visualized in: 2D planar view

Visualization in 3D

Rank Building

Light red (86) Sultan Abdul Samad

Dark orange (79) Merdeka Square

Dark orange (79) Royal Club

Orange (72) Library

Orange (72) Church

Orange (72) Bank Rakyat

Orange (72) Bank Pertanian

Orange (72) HSBC Bank Building

Dark yellow (65) Bukit Aman building

Dark yellow (65) OCBC Building

Dark yellow (65) CIMB Bank Building

Yellow (58) Muamalat Building

Yellow (58) Daya Bumi Building

Yellow (58) DBKL Building

Light Green (30) Tradewind Building

Green (23) Bank Negara Building

Dark Green (16) KWSP building

Dark Green 2 (9) Maybank

Refined hotspots after applying sniping spot quality ratio

No Building Degree Ratio MCA Total Score1 Sultan Abdul Samad building 110 0.611 86 52.5462 Royal Club Building 63 0.35 79 27.653 Bangunan Bukit Aman 44 0.244 65 15.864 Merdeka Square 30 0.167 79 13.1935 Church 17 0.094 72 6.7686 OCBC Building 14 0.077 65 5.0057 Public library 12 0.067 72 4.8248 Daya Bumi Building 12 0.067 58 3.8869 Bangunan Pertanian 9 0.05 72 3.610 DBKL Building 10 0.055 58 3.1911 Bank Rakyat Building 8 0.044 72 3.16812 CIMB Building 7 0.039 65 2.53513 Muamalat Building 7 0.039 58 2.26214 HSBC Building 4 0.022 72 1.58415 Tradewind building 7 0.039 30 1.1716 Bank Negara 5 0.027 23 0.62117 KWSP building 3 0.017 16 0.27218 Maybank 3 0.017 9 0.153

Rain effects

Rain effects: heavy

Control

Simulated for desert eagle

2D as buffer!

Rain effects: heavy; simulated for desert eagle and M16A

Rain effects for Draganov (800m)

Light

medium

heavy

In-situ verifications

Visibility analysis of 18 security hotspots

(a) Sultan Abdul Samad Building (100%)

(d) Public Library (38%) (c) Church (29.17%)

(d) Royal Club Building

(86.84%)

Assessments

randomly selected a total of 30 check points within the generated model

RMSE for planimetric and vertical accuracies are +1.129 and +1.288m, respectively.

Overall accuracy (RMSE) is +1.713m

Comparable to LiDAR generated DEM (Sun et al 2008; Muane 2001)

Conclusion

Demonstrated the use of integrated fine satellite remote sensing with airborne LiDAR data for identifying possible security hotspots in context of snipping from vantage points;

3D urban city block offers very effective visualization for security hotspot analysis;

Method demonstrated offers more practicality appeals to security industry and related decision makers;

Contribute to upcoming trend in the uses of spatial technology in security analysis in many aspects of public safety and homeland security.