Docstoc

Color

Document Sample
Color Powered By Docstoc
					 Color
Kyongil Yoon




               1
                                  Color
   Chapter 6, “Computer Vision: A Modern Approach”
   The experience of colour
            Caused by the vision system responding differently to
             different wavelengths of light.
   Radiometric vocabulary to describe energy arriving in
    different quantities at different wavelengths
   Human color perception
   Different ways of describing colors


2004-02-12                            VISA                           2
                       The Physics of Color
   Per unit wavelength to yield spectral units
            BRDF or albedo with wavelength
            Spectral radiance
   The color of source
            Black body radiators
                 Spectral power distribution depends only on the temperature of the body
                 Color temperature of a light source
            The sun and the sky
                 The sun: a point light source, daylight, yellow
                 The sky: a source consisting of a hemisphere with constant existence,
                  skylight (airlight), blue
            Artificial Illumination
                 Incandescent light: roughly black-body model
                 Fluorescent light: bluish tinge, mimic natural daylight
                 Others


2004-02-12                                      VISA                                        3
              The Physics of Color
   The color of surfaces
       Result of various mechanisms: different absorbtion
        at different wavelengths, refraction, diffraction,
        bulk scattering
       (Spectral) reflectance + (spectral) albedo

       Specular reflection




2004-02-12                   VISA                        4
                   Human Color Perception
   Color matching
       Let people match a given color using a certain number of primaries
       Subtractive matching
       Trichromacy
                Three primaries are required
                Subtractive matching, Independent
                Implies three distinct types of color transducer in the eye
       Grassman’s law
                If we mix two test lights, then mixing the matches will match the result
                 if    Ta = wa1P1+wa2P2+wa3P3 and Tb = wb1P1+wb2P2+wb3P3
                 then (Ta + Tb) = (wa1+wb1)P1+(wa2+wb2)P2+(wa3+wb3)P3
                If two test lights can be matched with the same set of weights, then they will
                 match each other
                 if    Ta = wa1P1+wa2P2+wa3P3 and Tb = wb1P1+wb2P2+wb3P3
                 then Ta = Tb
                Matching is linear
                 if    Ta = wa1P1+wa2P2+wa3P3
                 then kTa = (kwa1)P1+(kwa2)P2+(kwa3)P3
       Some exceptions
2004-02-12                                       VISA                                         5
                  Human Color Perception
   Color receptors
            We can assume that there are three distinct types
             of receptor in the eye that mediate color perception
                 Turns incident light into neural signals
            The principle of univariance
                 The activity of receptors is of one kind
            Rods and Cones
               Cones dominate color vision
               Three type of cones differentiated by their sensitivity
               S, M, and L cones (not necessarily blue, green, and red)


2004-02-12                              VISA                               6
                        Representing Color
                       (Linear Color Spaces)
   Linear color space
            Agree on a standard set of primaries
            Describe any color light by the three weights
   Easy to use
   Color matching functions
            Unit radiance source           U ( ) f1 ( ) P  f 2 ( ) P2  f3 ( ) P3
                                                             1
            Spectral radiance source       S ( ) w1 P  w2 P2  w3 P3
                                                        1
   How to deal with subtractive matching
            Negative weight value
   Standardization by CIE
            Commission international d’eclairage

2004-02-12                                 VISA                                            7
2004-02-12   VISA   8
              Linear Color Spaces
                   CIE XYZ
   Popular standard
   Color matching functions were chosen to be
    everywhere positive
   Impossible to
    get primaries




2004-02-12                 VISA                  9
                        CIE xy
   The horseshoe line
    (spectral locus) is the
    spectral locus.
   Hue changes one moves
    around the spectral locus
   Out-of-date?




2004-02-12                  VISA   10
                       Linear Color Spaces
   RGB
            Uses single wavelength primaries
             (645.16nm for R, 526.32nm for G, 444.44nm for B)
   CMY and black
            Red, yellow, blue: primary colors in subtractive mixture
            Simplest color space for subtractive matching
            Cyan (W-R), Magenta (W-G), Yellow (W-B)
            C+M = (W-R) + (W-G) = R+G+B-R-G = B
            Practical printer uses an additional black
                 Quality
                 Cost


2004-02-12                            VISA                              11
2004-02-12   VISA   12
                     Representing Color
                  (Non-Linear Color Space)
   Disadvantage of linear space
            Does not encode common properties such hue, saturation
            Not intuitive
   Hue, saturation, and value
            Hue: the property that varies in passing from red to green
            Saturation: the property that varies in passing from red to
             pink
            Value: brightness (lightness)




2004-02-12                             VISA                                13
2004-02-12   VISA   14
                     Non-Linear Color Space
                      Uniform Color Space
   Uniform color space
            The distance in coordinate space is a fair guide to the
             significance of the difference
            Just noticeable differences
                                                                 1
            CIE u’v’ space                                  Y 3
                                                          L*  116    16
                                4X           9X                   Yn 
             (u ', v ')               ,             
                           X  15Y  3Z X  15Y  3Z                     1      1
                                                                                     
                                                                      X 3  Y 3 
            CIE LAB                                      a*  500          
                                                                    X n   Yn  
                                                                             
                 Most popular                                     
                                                                                    
                 Good guide to understand how
                                                                         1       1
                                                                                    
                  different two colors will look                      Y 3  Z 3 
                  to a human observer                     b*  200     
                                                                    Yn   Z n  
                                                                   
                                                                                   
                                                                                    
2004-02-12                                  VISA                                   15
2004-02-12   VISA   16
             Spatial and Temporal Effects
   Chromatic adaptation
   Assimilation
   Contrast




2004-02-12                VISA              17
Statistical Modeling of Colour Data
   Daniel C. Alexander, Bernard Buxton
   Become standard to model
            Single mode distribution of color data by ignoring
             the intensity component and constructing a
             Gaussian model of the chromaticity




2004-02-12                        VISA                            18

				
DOCUMENT INFO