Lecture 4. Opponent Colors. Hue Cancellation Experiment HUV Color Space

Transcription

1 Lecture 4 Opponent Colors Hue Cancellation Experiment HUV Color Space

2 Opponent Colors Ewald Hering (1905) - Pure colors R G B Y. No such colors greenish-red, yellowish-blue Boynton & Gordon (1965) - With R G B Y can categorize all visible hues. Jameson & Hurvich (1955, 1957) - Hue Cancellation Experiments

3 Hue Cancellation Experiment test light canceling light Cancel the red-green content of the test light. Cancel the blue-yellow content of the test light.

4 Hue Cancellation Experiment

5 Hue Cancellation Hurvich & Jameson (1957) Red + Green cancellation lights Blue + Yellow cancellation lights

6 Hue Cancellation Hurvich & Jameson (1955) Unique Hues : yellow 573 nm, blue 472 nm, green 492 nm. Unique Red has some yellow (scarlet) Figure From

7 Physiological basis for Opponent Colors Svactchin & MacNichol ( 58) - Horizontal cells Boynton ( 79), DeMonasterio ( 78) - ganglion cells DeValois & DeValois ( 75) - LGN cells Derrington et al ( 84) - LGN cells rods cones horizontal ganglion light LGN (Lateral Geniculate Nucleus)

8 Physiological basis for Opponent Colors Opponent signals measured in LGN neurons L-M L+M-S Wavelength (nm) Derrington (1984)

9 Opponent process - possible neural connections: S M L L+M-S L-M L+M+S Blue-Yellow Red-Green Black-White Ganglion cells / LGN cells Cortical cells B- Y+ Y- B+ R- G+ B- R- G+ G- R+ Y+ Color Contrast detectors Color edge detectors

10 Opponent Cell - Neural Response G- R+ +R-G LGN cells Cell response + -

11 Chromatic On/Off Cells in Monkey Retina +M -L Green On/Red OFF Electrophysiological recordings midget ganglion cell in the monkey retina

12 Opponent Cell - Neural Response G- R+ + Cell response -

13 Opponent Cell - Neural Response G- R+ Cell response + - G- R+ Cell response Cell response + -

14 Mach Bands in Opponent Color Space Mach bands for small changes in lightness (top) and hue (middle)

15

16

17 McCullough s Effect - interaction between color and form

18 McCullough s Effect - interaction between color and form

19

20 Why Opponent process? A: Efficient Encoding Cone Spectral Sensitivity 1 S M L Relative sensitivity L and M cone sensitivities are highly correlated Wavelength (nm) Cone responses to several Natural SPDs : M-cone absorption L-cone absorption S-cone absorption M-cone absorption

21 Decorrelation: O 1 O 2 O 3 = L M S Spectral sensitivities of three decorrelated sensors 1 Blue-Yellow Red-Green Black-White (Decorrelated over the Macbeth color checker under mean daylight.)

22 B: Compression Human MTF MTF Spatial frequency Wavelength (nm)

23 Contrast Sensitivity

24 Contrast Sensitivity

25 Contrast Sensitivity Function Cambell Robson

26 Contrast Sensitivity Function Cambell Robson

27 Color Contrast Sensitivity

28 Asymmetric color matching experiment: test match (Poirson and Wandell 1993) Opponent channels have different modulations: 2 Spectral Sensitivities Contrast sensitivity Spatial frequency (cpd) Black- White Red- Green Blue- Yellow Wavelength (nm)

29 YIQ - Color Space NTSC = National Television Systems Committee Y I Q = X Y Z Y = luminance I = red-green Q = blue-yellow

30 YIQ - Color Space Target display image is RGB, derived from camera Y I Q For transmission in the US, the image is converted into YIQ Y = R G B I = R G B Q = R G B

31 YIQ - Color Space Target display image is RGB, derived from camera Y I Q

32 YIQ - Color Space Original Y - Blur I - Blur Q - Blur

33 Polar vs Opponent Color Spaces Opponent coordinates Polar coordinates

34 Polar vs Opponent Color Spaces Opponent coordinates Polar coordinates

35 Linear Color Spaces - Conversions Conversion matrix... From To XYZ LMS RGB OPP YIQ XYZ LMS RGB OPP YIQ Y I Q = X Y Z

36 Linear Color Spaces - Conversions Conversion matrix... M XYZ2YIQ = M XYZ2OPP = M LMS2OPP = M RGB2XYZ = M RGB2LMS = M XYZ2LMS =

37 Color Appearance Whether modeled in XYZ Color Space or HSV Color Space YIQ Color Space Color Appearance - is much more complicated!

38 Color Differences are non Uniform

39 Color appearance is context dependent All squares are matched on hue and chroma From: color/water_color/color3.html

40 Simultaneous Contrast

41 Lateral Inhibition

42 Lateral Inhibition

43 Simultaneous Contrast Apparent color shifts in simultaneous color contrasts Shifts are shown in contrast to a middle red color From: color/water_color/color3.html

44 Simultaneous Contrast color shift in a simultaneous chroma contrast all squares are matched on hue and lightness color shift in a simultaneous hue contrast all squares are matched on chroma and lightness color shift in a simultaneous hue contrast central squares set to low chroma and mid value; outer squares are both at lightness of 91 and 100% chroma

45 Simultaneous Contrast Boundary Effects

46 Simultaneous Contrast Boundary Effects

47 Simultaneous Contrast vs White s Illusion Simultaneous Contrast White s illusion

48 Color Induction

49 Color Appearance Color Appearance Depends On The Spatial Pattern (Spatial Frequency + Surround Color) Across The Cone Mosaic (Shevell and Monnier)

50 Color Appearance Immediate surround is not the main effect.

51 Monochromatic Effects Vasarely effect

52 Monochromatic Effects Vasarely effect From: Michael s Optical Illusions & Visual Phenomena

53 Chromatic Effects Vasarely effect Hues change steps along R-G

54 Craik-O Brien-Cornsweet Effect

55 Spatial Sensitivity Varies With Mean Luminance Level bright dark

56 Spatial Sensitivity Varies From Fovea To Periphery

57 Low, but not High,Temporal Frequency Sensitivity Varies With Mean Level a

58 Low Temporal Frequency Sensitivity Varies With Mean Level High Temporal Frequency does Not a

59 Measuring Color Differences???

60