Title: Chapter 6 Color image processing
1Chapter 6 Color image processing
- 6.1 Color Fundamental
- 6.2 Color models
- 6.3 Pseudo color processing
- 6.4 Basics of full-color image processing
- 6.5 Color transformation
- 6.6 Smoothing and sharpening
2Chapter 6 Color Image Processing
- Two major areas full color and pseudo color
- 6.1 Color fundamentals
- Primary color of light R, G, B
- Secondary color of light magenta, cyan and
yellow - Pigment or colorant subtract or absorbs a
primary color - of light and reflect or transmit the other two
- HIS model (a color may be characterized by its
chromaticity and brightness) - Chromaticity Hue and saturation taken together
- Hue is a dominant color as perceived by a
observer - Saturation is the relative purity or the amount
of white light - mixed with a hue
3Chapter 6 Color Image Processing
4Chapter 6 Color Image Processing
5Chapter 6 Color Image Processing
6- Brightness embody the chromatic notion of
intensity - Three basic properties used to describe the
quality - of chromatic light source source, radiance, and
brightness - Tri-stimulus values X, Y and Z
- Tri-chromatic coefficients
- CIE chromaticity diagram
- Has superior performance over other color
transforms especially in clustering of color
distribution and estimate of color difference - Shows color as a function of x (red) and y
(green) - Useful for color mixing
- Boundary of the diagram shows fully saturated
- Equal energy point
- CIE color models include CIE XYZ, CIE x,yY, CILE
Lab, and CIE Luv
7Chapter 6 Color Image Processing
8Chapter 6 Color Image Processing
96.2 Color models (Color space or color system)
- Purpose to facilitate the specification of
colors in some standard - A specification of a coordinate system and a
subspace within that system where each color is
represented a single point - Two applied directions for color models hardware
and applications where color manipulation (color
graphics) - RGB models--color monitors
- CMY (CMYK)--color model for color printing
- YIQcolor model for color television
- HIS a color model for humans to describe and to
interpret color decouple the color and
gray-level information
10- The RGB Color model
- Have a depth of 24bits
- The CMY color model
- The YIQ model
- The HIS color model
- Hue a color attributes that describes a pure
- Saturation a measure of the degree to which a a
pure color is diluted by a color - Brightness-- is a subjective descriptor that is
practically impossible to measure 0 - HIS is an ideal tool for developing image
processing algorithm
11Chapter 6 Color Image Processing
12Chapter 6 Color Image Processing
13- Converting colors from RGB to HIS
- Converting colors from HIS to RGB
- multiply H by 360o
- (1) When H is in RG sector (0? H ?120), the RGB
components are given by (Eq. 6.2-5 6.2-7)
14- (2) H is in GB sector (120? H ?240) Eq. 6.2-8
6.2-11 - (3) H is in BR sector (240? H ?360) Eq. 6.2-12
6.2-15 - Manipulating HIS component images
- Change the individual color of any region in the
RGB image - 6.3 Pseudo color (False color) image processing
- Differentiate the process of assigning colors to
monochrome images from the process associated
with true color images - Assign colors to monochrome images from the
process associated with true color images - To illustrate the power to allow independent
control over HIS (Fig. 6.17)
15Chapter 6 Color Image Processing
16Chapter 6 Color Image Processing
17Chapter 6 Color Image Processing
18Chapter 6 Color Image Processing
19Chapter 6 Color Image Processing
20Chapter 6 Color Image Processing
21Chapter 6 Color Image Processing
226.3.1 Intensity slicing (Color coding)
- If an image is interpreted as a 3-D function, the
method can be viewed as one of placing planes
parallel to the coordinate plane of the image - Each plane slices the function in the area of
intersection - Two color images whose relative appearance can be
controlled by moving the slicing plane up and
down the gray-level axis - Gray-level to color assignments are made
according to the relation f(x,y) ck if
f(x,y) ?? Vk (P planes, P1 intervals) - Plane is useful for geometric interpretation of
the intensity slicing technique
23Chapter 6 Color Image Processing
24Chapter 6 Color Image Processing
25Chapter 6 Color Image Processing
- The gray scale was divided into intervals and a
different color was assigned to each region (Fig.
6.20) - Is simple but powerful aid in visualization,
especially if numerous images are involved
26Chapter 6 Color Image Processing
27Chapter 6 Color Image Processing
286.3.2 Gray level to color transformation
- Achieve a wide range of pseudo color enhancement
- Perform three independent transformations on the
gray levels of any input pixels these results
are fed separately into the red, green, and blue
channels of a color television monitor. (Fig.
6.23) - Can be based on smooth, nonlinear functions,
which, as might be expected (based on a single
monochrome image) - Piecewise linear function (Fig. 6.19)
- Obtain various degrees of enhancement (Fig.6.24)
- Changing the phase and frequency of each sinusoid
can emphasize ranges in the gray scale
29Chapter 6 Color Image Processing
30Chapter 6 Color Image Processing
31Chapter 6 Color Image Processing
32Combine several monochrome images into a single
color composite (Fig . 6.26)
- Used for multi-spectral image processing
(different sensors produces individual monochrome
images) - Difference in color in various parts of the
Potomac River (Fig. 6.27a) - This processing help visualize events of interest
in complex images, especially when those events
beyond our normal sensing capabilities
33Chapter 6 Color Image Processing
34Chapter 6 Color Image Processing
35Chapter 6 Color Image Processing
366.4 Basics of full color image processing
- Two categories
- (1) Process each component individually and then
form a composite processed image from the
individually processed components - (2) Work with color pixels directly
- Color pixel are vectors
- Let C be an arbitrary vector
- Two conditions must be satisfied for
pre-component and vector-based - Has to be applicable to both vectors and scalars
- The operator must be independent of the other
components
376.5 Color transformations (Interactive
enhancement)
- Formulation of color transformation
- g(x,y)Tf(x,y)
- Color transformation (color mapping) of the form
si Ti r1 , r 2,,rn (n number of components) - For ex modify the intensity of the image on Fig.
6.30(a) g(x,y)kf(x,y) - In HIS color space, this can be done with s3kr3
- Only intensity component must be transformed
- Few operations but the conversion calculation are
more computationally intense
38Chapter 6 Color Image Processing
39Chapter 6 Color Image Processing
40- In RGB space, this can be done with three
components sik ri (i1,2,3.) - In CMYK space, it requires a set of linear
transformations (6.5-6) - Converting between representations must be
factored into the decision regarding the color
space in which to implement - Color complements
- Hue directly opposite one another on the color
circle - Analogous to the gray-level negatives
- Useful for enhancing details that is embedded in
dark region of a color image
41Chapter 6 Color Image Processing
42Chapter 6 Color Image Processing
43Chapter 6 Color Image Processing
446.5.3 Color slicing
- Highlighting a specific range of colors two
approaches - display the colors of interest
- Use the region defined by the colors as a mask
for further processing - Require each pixels transformed color components
to be a function of all n original pixels
components - slice a color image map the colors outside
some range of interest to a non-prominent neutral
color
45- Method 1 enclosed by a cube (a cube with width
W and centered at a prototypical color with
components) - Highlight the colors around the prototype by
forcing all other colors to the midpoint of the
reference color space - Method2 enclosed by a sphere
466.5.4 Tone and color correction
- Turn a PC into a digital darkroom
- The effectiveness of the transformations is
judged ultimately in print - Transformations are developed, refined, and
evaluated on monitors - Maintain a high degree of color consistency
between the monitor and the eventual output
device (device independent color model that
relates the color gamuts)
47- The success is a function of the quality of the
color profiles used to map each device to the
model and the model itself (CIE Lab model)- - Lab is useful for color management system (CMS)
- Lab is color metric (color perceived as
matching are encoded identically), perceptually
uniform (color differences among various hues are
perceived uniformly) and device independent
48- Its gamut encompasses the entire visible
spectrum, and can represent accurately the colors
of any display, print, or input device - useful in manipulation and image compression
applications - Lab is an excellent de-coupler of intensity and
color - The principal benefit of a calibrated imaging
system - allow tonal and color imbalances to be corrected
interactively and independently
49Corrections of Tonal ranges
- Key type (tonal range of an image)-refers to its
general distribution of color intensities - High-key image is concentrated at high
intensities - Low-key image are located at low intensities
- It is desirable to distribute the intensities of
a color image equally between the highlight and
shadow area - Modifying tones normally are selected
interactively - Operation adjust the images brightness and
contrast over a suitable range of intensities - the colors are not changed.
- In RGB and CMYK map all three components
- In HIS only intensity component is modified
50Chapter 6 Color Image Processing
51Chapter 6 Color Image Processing
52Chapter 6 Color Image Processing
53- Color imbalanceanalyzing with color spectrometer
- color wheel can be used to predict how one
component will affect another - The perception of one color is affected by its
surrounding colors - The portions of any color can be increased by
decreasing the amount of the opposite - Transformations are the functions required for
correcting the images (Fig. 6.36)
546.5.5 Histogram processing
- The gray-level processing can be applied to color
images in an automatic way - Produce an image with a uniform histogram of
intensity values - To histogram equalize the components of a color
image individually is unwise (results in
erroneous color) - A more logical approach spread the color
intensities uniformly (use HIS space), leaving
the colors themselves unchanged
55Chapter 6 Color Image Processing
566.6 smoothing and sharpening
- Modify values based on the characteristics of the
surrounding (without regard to its neighbors in
the previous section) - Instead of scalar gray-level values deal with
vectors - 6.6.1 Color image smoothing (Eq. 6.6-1 and 6.6-2)
- can be carried out on a per-color-plane basis
- smooth only the intensity component of the HIS
representation and convert the processed result
to an RGB image for display - 6.6.2 Color image sharpening
- Laplacian of vector c in RGB, and in I Component
of HSI
57Chapter 6 Color Image Processing
58Chapter 6 Color Image Processing
59Chapter 6 Color Image Processing
60Chapter 6 Color Image Processing
616.7 Color segmentationsegment objects of a
specified color range
- 6.7.1 Segmentation in HIS color space
- Carry out the segmentation process on individual
planes - color is conveniently represented in the hue
image - saturation is used as a masking image
- intensity is used less frequently for
segmentation (carry no color information) - 6.72. Segmentation in RGB vector space
- Better results generally are obtained
- Obtain an estimate of the average color vector
a - Classify each RGB pixel in a given image as
having a color in the specified range or not
have a measure of similarity (Euclidean distance)
62Chapter 6 Color Image Processing
63- D(z,a) z-a
- Distance measure with the covariance matrix
- D(z,a) D0
- Implementing (6-.7-1 and 6.7-2) is
computationally expensive - Distance measure with bounding boxa compromise
- Dimension is proportional to the standard
deviation of the sample along each of the axis - standard deviation is calculated using sample
color data - Segment it by determining whether or not it is on
the surface or inside the boxsimpler
computationally when compared to a spherical or
elliptical enclosure - Fig6.44b Yield much more accurate results
compared to HIS (correspond much more closely
with what we would define as reddish
64Chapter 6 Color Image Processing
65Chapter 6 Color Image Processing
666.7.3 Color edge detection
- Computing the gradient on individual images and
then using the results to form a color image will
lead to erroneous results - Individual based and vector-space based computing
depend on accuracy or just detecting edges (Fig.
6.45) - Extend the concept of a gradient from a scalar
function to vector functions
67- The def. of a gradient based on a vector (is a
vector pointing in the direction of maximum rate
of change f at (x,y) - Let r, g, and b be the unit vector along the R,G,
and B axis of RGB color space -
and - The direction of maximum rate of change of C(x,y)
is given by - The value of the rate of changes at (x,y) in the
direction is given by - gxxu.u ,gyyv.v, gxyu.v
68Chapter 6 Color Image Processing
69Chapter 6 Color Image Processing
70Chapter 6 Color Image Processing
716.8 The noise in color images
- Different noise levels are more likely to be
caused by differences in the relative strength of
illumination available to each if the color
channels - CCD sensors are noisier at low level of
illumination - Intensity component is slightly smoother than any
of the three noisy images (because intensity
image is the average of the RGB images)
72Chapter 6 Color Image Processing
73Chapter 6 Color Image Processing
74Chapter 6 Color Image Processing
756.9 Color image compression
- The data that are the object of any compression
are the components of each color pixel - Compression is the process of reducing and
eliminating redundant and/or irrelevant data
76Chapter 6 Color Image Processing