Learning Today, Leading Tomorrow Introduction to Color Science & Engineering 울산과학기술대학교 디자인 및 인간공학부 곽영신
Contents Introduction What is Color Science? What is Color Engineering? Research Topics Color Appearance Modeling Memory Color Reproduction RGB-to-RGBW color conversion technique for EPD Color Science & Engineering Lab @ UNIST
What is Color Science?
What is Color Science?
What is Color Science? Colorimetry Color/Image Appearance Model Uniform Color Space Color Difference Equation Visual Appearance Model Color/Image/Lighting Quality, Preference
What is Color Engineering? Color Imaging Industry Typical image reproduction in office environment Reproduction Original Image Reproduction Typical image reproduction at home 방송국 방송 싞호
What is Color Engineering? Paint Industry Instrumental-based Color Matching Colorant formulation prediction
COLOR APPEARANCE MODELING
Color appearance model To predict changes of color appearance of a visual stimulus under various different viewing conditions. Color Appearance data Psychophysical Experiments using magnitude estimation method Observers were asked to estimate the lightness, colorfulness and hue of each test color Structure of Color Appearance Model XYZ in the test condition Background, surround condition Chromatic Adaptation Dynamic Adaptation Reference Colorfulness Background Test Color Reference White Opponent Color Signal Colour Appearance Predictors Lightness, Brightness Saturation, Chroma, Colorfulness Hue
Input Data and Parameters Relative tristimulus values of test stimulus X Y Z Relative tristimulus values of the reference white Xw Yw Zw Reference white in reference conditions Xwr=100 Ywr=100 Zwr=100 Photopic luminance of the adapting field (normally taken as 1/5 of the lumininace of reference white) LA (cd/m2) Background luminance factor Surround parameters F c Nc Background parameters Ncb Nbb Yb Surround Conditions F c Nc Average Surround 1 0.69 1 Dim surround 0.9 0.59 0.95 Dark surround 0.8 0.525 0.8
Chromatic Adaptation A modified CMCCAT2000, CAT02 3.6 1 1 1, 1, 1 0 9834 0 0136 0 0030 0 0061 16975 0 7036 01624 0 4296 0 7328 where 1 02 92 42 02 02 C C C CAT C C C L W WR C W WR C W WR C CAT CAT B G R M Z Y X e F D B D B B D B G D G G D G R D R R D R......... M Z Y X M B G R A
Log10(FL) Log10(Signal after Dynamic Adptation) Dynamic Adaptation R, G B : Hunt-Pointer-Estevez cone space R G B ' ' ' M H X Y Z C C C where M H 0. 38971 0. 68898 0. 07868 0. 22981 118340. 0. 04641 0. 00000 0. 00000 100000. 6 4 2 0 Dynamic fn. CIECAM02 Power fn. R B ' a ' a 400 400 where L F L 0.2k 0.42 ' FL R 100 0.1, ' 0.42 FL R 100 27.3 ' 0.42 FL B 100 0.1 ' 0.42 F B 100 27.3 4 G ' a 400 4 2 1/3 5L 0.11 k 5L, k 1 5L 1 A F L G A F L ' 0.42 ' G 100 0.1 0.42 100 27.3 A -2-10 0 10 Log 10 (F L R'/100) 4 2 0-2 -4-6 -8-10 0 10 Log 10 (5L A )
Cone Sensitivity after Compression (CIECAM02) Achromatic Sensitivity (CIECAM02) Opponent Color Signals Achromic Redness - Greenness Yellowness Signal A - Blueness 2 R a b ' a G R 1 9 ' a ' a R ' a 0.05 B 12 11 G G ' a ' a ' a 1 11 2 B 0.305 B ' a ' a N bb where N bb 0.725 n 0.2, n Y Y b w 15 10 B'a G'a R'a 1.2 0.8 A V () 5 0.4 0 350 600 850 Wavelength (nm) 0.0 350 600 850 Wavelength (nm)
Color Appearance Predictors Achromatic Predictors Lightness : J, Brightness: Q 25 0. 0.5 4 100 c 4 1.48, 100 L W w b cz w F A J Q Y Y z A A J
Color Appearance Predictors Chromatic Predictors Chroma : C, Colorfulness: M, Saturation: s C t 0.9 J 100 1.64 0.29 n 0.73 where t R ' a e G ' a 2 2 a b 21 20 B ' a, N cb Y 0.725 Y w b 0.2 e 12500 13 N c N cb cos h 180 2 3.8 M C F 0.25 L s 100 M Q 0. 5
Color Appearance Predictors Hue Predictors Hue angle : h, Hue quadrature: H h tan 1 b a [degrees] H H 1 100h h1 e1 h h1 e1 h2 h e2 Unique Hue Red Yellow Green Blue Hue angle h 20.14 90.00 164.25 237.53 Eccentricity e 0.8 0.7 1.0 1.2
Mechanisms of Adaptation
Effect of Luminance Level
Simultaneous contrast White위의 색상들은 black위의 색상에 비해 더 어둡고 작아 보인다 (Lightness induction) Illusion due to relative lightness
Size Effect
Spatial Contrast Sensitivity
MEMORY COLOR REPRODUCTION
Preferred Color Reproduction Image quality on TV The reproduced image on TV and original scenes are seldom observed side by side. The naturalness of reproduced image is positively correlated with image quality and the naturalness of a whole picture is determined by the naturalness of the memory colors such as skin color, sky, grass, etc. in that picture Therefore, properly reproduced memory colors in an image can improve image quality regardless the color reproduction of other area
y Preferred Color Reproduction Psychophysical Experiment Ten test images were used in the experiment and each image was transformed using memory color transformation algorithm to have different nine green grass / red colors. The pair comparison method was used to find out the most preferred green grass / red colors. Toggle Select 0.60 0.50 0.40 Y Input Target After Chromatic Transformation Green Grass 0.30 0.20 0.30 0.40 0.50 x Cb, Cr
CIECAM02 hue angle (h) CIECAM02 hue angle (h) CIECAM02 hue angle (h) Preferred Green Grass Colors Preferred color was analyzed using CIECAM02 Preferred Green Grass Color Having the highest chroma with the hue closest to the hue angle 130 i.e. around 165 (65% green and 35% yellow) in terms of hue quadrature. 180 180 180 img 5 160 140 img 8 img 4 160 140 img 10 160 140 img 7 img 6 120 img 3 100 10 20 30 40 50 60 70 CIECAM02 Chroma (C) img 9 120 img 1 100 10 20 30 40 50 60 70 CIECAM02 Chroma (C) 120 img 2 100 10 20 30 40 50 60 70 CIECAM02 Chroma (C) CIECAM02 chroma and hue angle of the average green grass colors of the test images (filled symbol: the most preferred image)
CIECAM02 hue angle (h) CIECAM02 hue angle (h) Preferred Accent Color vs. Memory Color Preferred Apple Color The most preferred red hue for each image is directing the unique red i.e. hue angle 20.14 / Preferred Max. Chroma: near 75 Preferred Red Object Color Highest Chroma is preferred regardless hue 40 40 30 img 6 img 3 img5 30 img10 20 img 4 20 img 8 img 7 10 img 1 img 2 10 img 9 0 20 40 60 80 100 CIECAM02 Chroma (C) 0 20 40 60 80 100 CIECAM02 Chroma (C) Preferred Apple Colors Preferred Red Object Colors
RGB-TO-RGBW COLOR CONVERSION TECHNIQUE FOR EPD
RGBW color techniques Why RGBW 4 Color? More Brightness with no extra cost. The most effect luminance increment method for the displays with backlights such as LCD Technical Issues New pixel structure Color signal conversion algorithm (RGB RGBW) Increase luminance without decreasing colorfulness
CIELAB L* CIELAB b* CIELAB L* RGB vs. RGBW 250 100 80 RGBW RGB 60 0 40 RGBW RGB -250-100 -50 0 50 100 CIELAB a* 20 0-100 -50 0 50 100 CIELAB a* 100 80 RGBW RGB 60 40 20 0-100 -50 0 50 100 CIELAB b*
Working mechanism of EPD Microencapsulated Electrophoretic Display (EPD) 젂기이동(Electrophoresis) 현상을 이용한 반사형 디스플레이 Ink-on-paper appearance, Low power consumption, Flexibility Color EPD 상업화를 위하여 낮은 반사율, 낮은 contrast ratio 및 color gamut 극복 필요 EPD 구조 및 구동 방법 TFT layer에 걸리는 젂압 크기와 시갂에 따라 gray 계조 표현 - - + + 0 V Data Voltage IZO layer
y y Reflectance Color Measurement EPD Full white 반사율: 12.8% (싞문지 반사율: 45~55%) 20% L* a* b* X Y Z 16% R G 12% B W 8% Black Full W 4% 0% 380 430 480 530 580 630 680 730 Wavelength (nm) Color gamut size in xy-plane: 3.7% of NTSC Black 20.81-2.27-0.57 2.91 3.20 3.57 Red 26.13 5.81 3.83 5.01 4.79 4.44 Greed 28.86-13.46 7.29 4.43 5.78 4.68 Blue 24.38-2.07-9.15 3.87 4.22 6.65 White 30.81-1.87-1.10 6.07 6.57 7.45 Full W 44.53-4.07 1.26 12.88 14.21 14.92 At least 6-colors (RGBCMY) are needed to define color 0.90 0.60 0.30 0.00 B G R 0.00 0.20 0.40 0.60 0.80 x EPD srgb NTSC 0.42 0.38 0.34 0.30 0.26 G Y C R M B 0.24 0.27 0.30 0.33 0.36 x
y y Crosstalk between sub-pixels R G R G R G R G R G R G R G R G R G R G R G R G R G R G R G R G R G R G W B W B W B W B W B W B W B W B W B W B W B W B W B W B W B W B W B W B R G R G R G R G R G R G R G R G R G R G R G R G R G R G R G R G R G R G W B W B W B W B W B W B W B W B W B W B W B W B W B W B W B W B W B W B R G R G R G R G R G R G R G R G R G R G R G R G R G R G R G R G R G R G W B W B W B W B W B W B W B W B W B W B W B W B W B W B W B W B W B W B Adjacent (Colors) Adjacent (W+Color) Diagonal Two colors Three colors Y increment C* increment Y increment Adjacent (Colors) (R+G,G+B) 7.0% 14.8% 15.0% Adjacent (W+Color) (R+W,B+W) 12.4% 18.6% 20.2% Diagonal (R+B, G+W) 15.4% -3.5% 0.42 EPD Gamut 0.42 EPD Gamut 0.38 0.34 Measured Added G+B G+W R+W R+G 0.38 0.34 Measured Added G+B G+W R+G R+W 0.30 B+W B+R 0.30 B+W B+R Max Color (255) 0.26 0.24 0.28 0.32 0.36 x Mid Color (128) 0.26 0.24 0.28 0.32 0.36 x
Fringe Field & Scattering Effect of Fringe field + 0 V Fringe Field + 0 V 연달아 있는 두개의 부화소가 켜진 경우 fringe field 영향 감소로 유효 반사 면적 각각을 켰을 때 측정된 색값의 합보다 더 밝 고 진한 색을 보임 Effect of Scattering 0 V 0 V 0 V 0 V Color 부화소들끼리 연달아 있는 경우보 다 color와 white 부화소가 켜진 경우 부 화소 사이의 산란 현상 때문에 채도 향상 율이 더 높아짐
RGB-to-RGBW Signal Decomposition EPD용 RGB-to-RGBW signal decomposition algorithm Primary color 출력 시 white signal을 추가하여 반사율 향상 산란 현상에 의해 white 추가에 의한 채도 저하 미미 고채도 색상 표현 시 white를 이용한 채도 조젃 White 사용 후 다른 RGB 채널 값들을 활용한 채도 조젃 P P 남은 채널 값 추가 S th V th White만 추가 W add V enhance
Conclusion Spectrophotometer를 이용 RGBW EPD 패널의 색측정 결과 EPD color gamut 정의를 위해 최소 RGBCMY 6개 color 측정이 요구됨 부화소 갂에 심각한 crosstalk 현상으로 인해 기존 display의 color management 기술 적용이 어려우므로 EPD를 위한 새로운 기술 개발이 요구됨 EPD에서의 crosstalk은 부화소 사이에 형성되는 fringe field와 white ball 표면에서의 scattering 현상에 의해 발생 Fringe field 때문에 primary color 측정 시 유효 반사 면적이 감소되어 color gamut이 작아지는 원인이 됨 인접한 부화소가 같이 켜질 경우 fringe field의 영향을 받는 면이 감소되어 밝기 및 채도 증가 정도가 각각을 켠 경우를 합한 경우 보다 커지게 됨
Conclusion 부화소 사이에서 산란되는 빛의 영향으로 color 부화소가 white 부화소와 인접한 경우 white 추가에 의한 밝기 향상 효과는 커지며 채도 감소 정도는 작아짐 EPD 특성을 반영하여 화질을 최적화 하기 위한 RGB-to-RGBW signal decomposition algorithm을 개발함 순색에 white 추가 및 white만을 이용한 채도 조젃로 고채도 색상들의 밝기 향상 scattering 현상에 의해 고채도 색상의 채도 감소 최소화 개발된 알고리즘 성능 평가 결과 기존 LCD 패널을 기준으로 개발된 알고리즘 대비 90% 이상의 경우 선호됨 EPD의 물리적 특성인 fringe field 및 산란 현상을 활용하여 개발된 색처리 알고리즘을 이용할 경우 EPD 영상 화질을 크게 향상 시킬 수 있음
COLOR SCIENCE & ENGINEERING LAB @ UNIST
Current Research Topics at Color Lab. Visual Appearance Color and Surface Appearance Model based on Pattern Analysis 광택 및 Pearl이 있는 표면색에 적용 가능한 색차식 연구 Image Quality Stereo 3D TV의 Image Quality Issues UDTV향 Color 화질 향상 기술 개발 Color & Design 색 감정 모델 연구 LED 조명 스펙트럼 구성이 조명 선호도에 미치는 영향 연구 U-VISIT (UNIST Open Lab) : 2010. 10. 2. (토)
Questions