Color Image Processing

Color Image Compression

 Color Image
 Color is a powerful descriptor that helps or simplifies object
identification, and extraction of objects from a scene.

 A color image is an M x N x 3 array of color pixels, where each

pixel is a triplet corresponding to the color components.

 Appropriate color models/color spaces according to the application

are used generally to facilitate the specification of colors in some
standard, accepted way.

Eg. RGB, HSI, YUV, HSV, CMY, CMYK, YCbCr, CIE L*a*b* ,
CIE L*u*v* etc.
 Color Perception
 3 types of cones: L (64%), M (32%), S (2%)
 Retinal Physiology and Color
 Human retinas have (at least) four types of
photoreceptors
 Three types of ‘cones’
 High light level, high acuity vision

 Each type of cone has a different spectral

response
 One type of ‘rods’
 Low-light level and peripheral vision
 Color Perception
 Tri chromatic theory
 Opponent Color theory
 HVS Color perception can be modeled in two steps
 The received light is first transformed into three signals as
received by the three types of cones in the retina.
 The received signals are combined into three components one
achromatic and two chromatic.

 Red/Green opponent signal: R-G

 Yellow/Blue opponent signal: (R+G)-B
 Color Perception
Original

Red-Green Blue-Yellow
opponency opponency
Primary and Secondary colors
Additive Color System: By
adding the primary colors of
light
Subtractive Color System: By
subtracting a primary color of
light, absorbing the primary
color and reflecting the other
two – eg. printing industry
Eg. Yellow – absorbs blue and
transmits Red & Green
 Color Specification
 Color can be specified using the tri-stimulus values R, G,
and B.

 The amounts of Red, Green and Blue needed to form a

particular color are called the tri-stimulus values.

 A color can be then specified by its tri-chromatic

coefficients, (x,y,z) where
R G B
x ,y ,z 
RG B RG B RG B
 Color can be specified in terms of Hue (H) Saturation (S)
and Intensity (I).
Chromaticity Diagram
Colour Gamut
 Color Models
 Specify colors in a standard accepted way.

 A color model is a specification of a coordinate system and

a subspace within that system where each color is
represented by a single point.

 Oriented toward hardware or toward application where

color manipulation is the goal.

 Eg. RGB, HSI, NTSC, HSV, CMY, CMYK, YCbCr, CIE

L*a*b* , CIE L*u*v* etc.
 RGB Color Model
 Specify Colour in terms of three primary colours: red, green
and blue, mixed together.

 The model is based on a Cartesian coordinate system

 Used in television, computer monitors, etc.

 We specify the levels of R, G and B in the range [0, 1], but

they can easily be extended to other ranges (8-bit integers for
example).

 A full colour image – 24-bit RGB. Safe RGB colours – 216

+ 40 – 256 colours
The RGB Color Cube
 HSI Color Model
 RGB is good for generating colors but not for
describing colors as humans perceive them.

 HSI - Specify color in terms of Hue Saturation and

Intensity.

 Color information is decoupled from intensity – I

gives an intensity image.
 RGB - HSI model
Colors on this triangle
Have the same hue

Intensity
line

saturation
RGB - HSI model
HSI model
 RGB to HSI conversion
 if B  G
H 
360   if B  G
with
 

1
( R  G )  ( R  B) 
  cos  1 2
1 



 ( R  G ) 2  ( R  B )(G  B ) 2

3
S  1 [min( R, G , B )]
( R  G  B)
1
I  ( R  G  B)
3
HUE BRIGHTNESS SATURATION

R
G
B

HUE BRIGHTNESS SATURATION

M
Y
C
 CIE L*a*b* Color Space

 The CIE L*a*b* Color Space

 L* defines lightness, a* denotes the red/green
value and b* the yellow/blue value
 The CIE L*a*b* is approximately perceptually
uniform
 There exists no direct conversion between the
RGB and the CIE L*a*b* color space
 CIE L*a*b* Color Space
 the conversion can be done via the CIE XYZ color space
 The linear transformation between RGB and XYZ values is
as follows  X  0.412453 0.357580 0.180423  R 
 Y    0.212671 0.715160 0.072169 G 
    
 Z  0.019334 0.119193 0.950227  B 

 The transformation between XYZ and L*a*b* values is

nonlinear:

116 .(Y / Yn )
1/ 3
 16 Y / Yn  0.008856
L 
*

903 .3.Y / Yn otherwise
a *  500 .( f ( X / X n )  f (Y / Yn ))
t 1/ 3 t  0.008856
b  200 .( f (Y / Yn )  f ( Z / Z n ))
* f (t )  
7.787.t  16 / 116 otherwise
 CIE L*a*b* Color Space

 Xn, Yn and Zn denote the tristimulus values of the

reference white.
 For D65, [Xn, Yn, Zn] = [95.047,100,108.883].
 The YCbCr Color Space

 The YCbCr Color Space

 YCbCr is a family of color spaces used as a part of the
Color image pipeline in video and digital photography
systems.
 Y is the luma component and Cb and Cr are the blue-
difference and red-difference chroma components.
 YCbCr is a practical approximation to color processing
and perceptual uniformity, where the Primary colours
corresponding roughly to Red, Green and Blue are
processed into perceptually meaningful information
 The YCbCr Color Space

 To convert from RGB to YCbCr, the

following operation is used.

Y  0.2989 R  0.5866G  0.1145B

Cr  0.1687 R  0.3312G  0.5000 B

Cb  0.5000 R  0.4183G  0.0816 B

Processing Color Images
 Introduction

 Two major areas of color image

processing are
 Full color Processing: Images acquired with
full color sensors are used for processing

 Pseudo Color Processing: A color will be

assigned to a particular monochrome intensity
or range of intensities.
 Pseudo Color Processing
Intensity slicing

 3-D view of intensity image

Color 1

Color 2

Image plane
 Alternative representation of intensity slicing
 More slicing plane, more colors
 Application 1

Radiation test pattern 8 color regions

* See the gradual gray-level changes

 Application 2

X-ray image of a weld

Application 3

Rainfall statistics –
in tropical regions of
earth
Washington D.C. – Potomac River

R G

Near
Infrared
(sensitive
B to biomass)

R+G+B near-infrared+G+B
 Full Color Processing
 Two approaches
 Each component processed separately –
Per-Color-Component Processing.

 Work with each pixel treating it as a

vector having three components – Vector
Based Processing.
 Color pixel

 A pixel at (x,y) is a vector in the color space

 RGB color space
 R( x, y) 
c( x, y)  G( x, y)
 B( x, y) 
 gray-scale image

f(x,y) = I(x,y)
Example: spatial mask
 How to deal with color vector?
 Per-color-component processing
 Process each color component

 Vector-based processing
 Process the color vector of each pixel

 When can the above methods be equivalent?

 Process can be applied to both scalars and vectors

 Operation on each component of a vector must be

independent of the other component

 Full Color Processing
Example Marginal  R  100 
  
median  V  100 
 
 B0 
 R  100   R0   R  100 
     
 V 0   V  100   V  100 
     
 B0   B0   B  100 
 R0 
 
Vectorial  V  100 
median  
 B  0 
Full Color Processing

False
Colours !!
Marginal
Median

Vectorial
Median
 Full Color Processing
 Color Transformations
 Intensity modification
 Color complements
 Color Slicing
 Histogram processing
 Color image smoothing and sharpening
 Noise removal
 Color image Compression
 Color image segmentation
 Color transformation

 Similar to gray scale transformation

 g(x,y)=T[f(x,y)]

 Color transformation
si  Ti (r1 , r2 ,..., rn ) , i  1,2,..., n
 Use which color model in color
transformation?
 RGB CMY(K)  HSI

 Theoretically, any transformation can be

performed in any color model

 Practically, some operations are better suited to

specific color model
 Example: modify intensity of a color
image
 Example: g(x,y)=k f(x,y), 0<k<1
 HSI color space
 Intensity: s3 = k r3

 RGB color space

 For each R,G,B component: si = k ri

 CMY color space

 For each C,M,Y component: si = k ri +(1-k)

Processing results are the same. Which operations is the fastest?

Original image k=0.7 (reduce intensity)

I H,S
 Color complement

 The Color Circle

Transformation: Color complement
 Color slicing

 Recall the pseudo-color intensity slicing

1-D intensity
 Color slicing
 How to take a region of colors of interest?

Sphere region Cube region

prototype color
prototype color
Application

cube sphere
 Color Slicing


0.5 if [ r j  a j  W 1 j  n
si   2

ri otherwise


if  rj  a j   R0
n
1 j  n
2 2
0.5
si   j 1
r
i otherwise
Histogram processing
 Color image smoothing

 Neighborhood processing
 Color image smoothing: averaging
mask
1
c( x, y ) 
K
 c( x, y )
( x , y )S xy
vector processing

Neighborhood
Centered at (x,y)

1 
  R ( x, y ) 
 K ( x , y )S xy 
1 
c( x, y )    G ( x, y )  per-component processing
 K ( x , y )S xy 
1 
 K  B ( x, y ) 
 ( x , y )S xy 
Image Smoothing
Image Smoothing
Image sharpening
Image Compression