spectral color imaging

SPECTRAL COLOR IMAGING

Jussi ParkkinenMarkku Hauta-Kasari

IPCV 2006August 23th, 2006

Budapest, Hungary

Color image formation in human eye

Reproduction of color images on displays

Display characterictics

Laptop, white color on display

350 400 450 500 550 600 650 700 750 8000

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6x 10

-3 valkoinen kannettava

Multiprimary color displays

Conventional LCD 4-primary Flat-panel LCD2x2-tiled, 2000x2000pixels rear-projection6-primary display

Stacked front-projection 6-primary DLP display

y

x

Visible Color for Human Eye

Color Gamut presented by Multiprimary color display

Color Gamut presented by RGB display system

RGB-filters

High and low pass filters

Multiprimary display

6 filters for

Literature• Wyszecki and Stiles: Color Science. John Wiley & Sons,

2000.– The “Bible” of Color Science. Not new anymore, but basics of

standard color science and basics of color are valid material.• Roy Berns: Billmeyer and Saltzman's principles of color

technology. John Wiley & Sons, New York (NY), 2000.– Good and clear introduction to the color and related technologies.

• Hardeberg: Acquisition and reproduction of color images : colorimetric and multispectral approaches. Dissertation.com, 2001 – Jon Yngve’s PhD thesis, good overview on spectral approach and

some methods there.• Mark Fairchild: Color Appearance Models. Addison

Wesley, Reading (MA), 1998.– Explains methods to be used for reproduction of colors to look

correct

SPECTRAL IMAGE APPLICATIONSJussi Parkkinen

Markku Hauta-Kasari

IPCV 2006August 23th, 2006

Budapest, Hungary

Applications areas of spectral imaging• Medical imaging, telemedicine*

• Cultural heritage study and digital museums*

• Paper industry*

• Printing industry*

• Textile industry• eCommerce• Plastic industry*

• Cosmetic industry• Display technology*

Testimage for printing quality test (newspaper)

Some sample spectra for original and print

Example: printing inks and measurement

Example: spectral measurements

Same ink in same paper(different amount of ink)

Comparison between pre-print and print

pre-print print

color difference

Mean reflectance spectra (Jacket)Mean whole image

Ref

lect

ance

print

pre-print

Examination of metameric inkCourtesy by Yoichi Miyake

Color image With long pass filter>645nm

Passport of Japan (Personal page)

Example of fluorescent colors

Examples: color characterization

Videoconferencing using web-cam

Webcam color correction

Original image Color calibrated image

Simulation of illuminant change

Spectral component images (400, 550,

700 nm)

Church

Observation of Spectral Reflectance

OriginalOnceTwiceThree Times

Example of co-operation with local industry: Tulikivi Co. Ltd. Crevice analysisImage optimization for a screen

IMAGE OF ORIGINAL SAMPLE

IMAGE OF SRGB-REPRODUCTION

ON CRT DISPLAY

IMAGE OF METAMERICCOLOR REPRODUCTION

ON CRT DISPLAY

Figure 3Figure 3. . Differences between original and metameric reproduction (especially lower and upper parts)

are because of uneven illumination of the original sample when picture was taken.

Figure 4Figure 4. . Original printed sample in the middle.

Spectral imaging of displays

On the left ImSpector spectral camera and a 45 degree angle mirror. On the right HP laptop display and Nokia cellular phone display.

Spectral images transformed into sRGB

Laptop LCD display Cellular phone display

Principal component analysis of spectral images

1-4

Laptop LCD display Cellular phone LCD display

5-8

9-12

1316

17 20

Acquiring of spectral face images

Part of the database

Spectral color enhancementapplied to “psoriasis vulgaris”

natural color Enhancement (550nm band)

• Skin lesion apparently visualized by spectral color enhancement

Creation of color surface coating for gray level objects

•Colors of cherries, red apples and a blue pen were picked and used for coloring the cylinder.

CCDcamera

ＡＯＴＦ

Lens

Halogen Lamp (pole)

Plants

Experimental setup

no_ozoneozone

Before ozone exposure （ 0 hours)

Slide14

ozone exp

visible RED 59 hours

ozone exp

4 hours

⊿NIR had remarkable change in parts where visible damage occurred

３

３invisible NIR

Measurement of soybeanEmission Wavelength (LCTF)

400nm 410nm 600nm

350n

m

Exc

itat

ion

Wav

elen

gth

(G

rati

ng

)

360n

m57

0nm

Total 273 imagesTotal 273 images

Excitation-Emission Matrix

SpectralIlluminator

Spectral Imager

XenonLamp

Grating

Micro-slicer

3D Spectral 3D Spectral Imaging SystemImaging System

CCD

LCTF

Measured at National Food Reserch Institute

Result

Proposed method

spectral image

digital camera image

Sugar Content Map(Assoc Prof S. Nakauchi Toyohashi U of Tech, Japan)

知覚画像 糖度分布可視化画像 BrixMelon0726_28- 1

7

891011121314

15

知覚画像 糖度分布可視化画像 BrixMelon0726_27

7

891011121314

15

Multispectral microscopic image capturing system16-band, 2k x 2k pixels / channel

16-band multispectral camera for still image

Multispectral cameras

6-band HDTV camera for motion picture

Spectral video

• Sequence of spectral images

• Shown as movie on the screen

• Very high memory requirements

• Efficient compression needed

• Shown as RGB or multiprimary

AcknowledgementResearchers in Joensuu Color Group

andDr. P. Laihanen Helsinki U of Tech, FinlandAssoc Prof Y. Manabe NAIST, JapanProf Y Miyake Chiba University, JapanDr. Miyata National Museum of Japan HistoryDr. K. Miyazawa Toyohashi U of Tech, JapanAssoc Prof S. Nakauchi Toyohashi U of Tech, JapanProf G. Nyman University of Helsinki, FinlandMr. E. Torniainen M-Real Company, FinlandAssoc Prof N. Tsumura Chiba University, JapanAssoc Prof M. Yamaguchi Natural Vision Center, Japan