Title: Lightness filtering in color images with respect to the gamut
1Lightness filtering in color images with respect
to the gamut
Fourteenth Color Imaging Conference Judith Dijk
and Piet W. Verbeek
Presented by Soo-Jin Sung
- School of Electrical Engineering and Computer
Science - Kyungpook National Univ.
2Abstract
- Proposed method
- A generic method that allows grey image
processing for lightness processing on color
images - without exceeding the limits of the gamut of the
technique or device - The results of the proposed method
- Sharpening improvement
- Contrast improvement
3Introduction
- A difference between grey images and color images
- Grey images
- Having a simple scale between black and white
- Color images
- The range of colors
- Proposed method
- A generic method
- Appling grey image processing on the luminance of
color images - Without exceeding the limits of the output gamut
4The generic method
- The base
- Image processing of color images
- Hue, chroma or saturation, and lightness
- The constraint
- The colors of the input image are located within
the gamut - Definition for colors as points in the CIELAB
space
5- The desired lightness correction
- The move way of the original color point along a
fixed path in a constant hue plane - The path
- The color of the original point ( position in the
gamut and the relation between the lightness and
chroma change) - The fixed hue of the point
- Keeping the same chroma before and after the
image processing - The prevention of the colorfulness of the image
Figure 1. Illustration of a path for a given
point x with chroma constant.
6Gamma manipulation
- The effect
- The lightness values are distributed nonlinearity
over the range - Contrast improvement
-
- More contrast in the higher colors, and decreased
mean lightness -
- More contrast in the darker colors, and increased
mean lightness
(1)
7Figure 2. Gamma manipulation.
Figure 3. Gamma manipulation, while keeping
chroma constant. (a) the normal, not
gamut-limited, gamma manipulation (b) the gamma
manipulation as defined in equation 1
8Recipes for gamut-limited gamma manipulation
- Possible recipes
-
- The manipulated lightness while keeping the
chroma constant -
- The manipulated lightness while keeping the ratio
constant - Mapping towards black and white
- Movement toward black for a lightness decrease
- Movement toward white for a lightness increase
- Mapping away from black and white
- Movement away from black for a lightness increase
- Movement away from white for a lightness decrease
9(a)
(b)
(d) Mapping away from black and white
(c) Mapping towards black and white
Figure 4. Paths along which a color point may
move within the constant hue plane, when applying
different recipes for gamma manipulations.
10Results for gamut-limited gamma manipulation
- In figure 5
- Clipped out-of-gamut pixels
- For low gamma values, too yellow in a skin color
- In figure 6
- Having the smaller effect, because the mean
lightness change is smaller - The more natural skin tones than the figure 5
Figure 5. The result for normal gamma manipulation
11- In figure 7
- chroma increase,
chroma decrease - Having the somewhat less quality than the images
in figure 6 - In figure 8
- mixed white,
mixed black - Less vivid and less than quality of other recipes
- In figure 9
- No difference very much from each other
Figure 8. The results for the mapping towards
black recipe.
Figure 9. The results for the away from black
recipe.
12Gamut limited sharpening
- Process of sharpening
- Changing the lightness of a point
- Adding a high frequency filtered version of the
image to the original image - Used unsharp masking
- The high frequency filtered version
- The unsharp masking sharpening filter
- Inverse unsharp masking filter
(2)
where the fourier transform
(3)
(4)
13Recipes for gamut-limited sharpening
- Tested recipes
- Sharpening within the gamut
- The maximum and minimum sharpening for each point
within the gamut - Gamut limited sharpening
- The desired lightness difference adjusted so that
the maximum lightness difference - Adjustment of the value of
- Mapping towards black and white
- Selection of the new point through the original
point and white or black - Mapping halfway towards black and white
- Mapping away from black and white
14Figure 10. Different paths for sharpening
manipulation.
15Results for gamut-limited sharpening
16- The sharpest images ((b) and (d))
- The effect on the location of the eyes and the
mouth - The more colorful image (b) than the image (d)
- The quite visible chroma change in the low
frequency for the (d) - (e) recipe
- Adding chroma on places where the lightness is
changed - More colorful than the colors in the original
image - In the colors of the mouth of the woman
17- (a) and (c) recipe and the normal sharpening
- Adjustment of the size of for these recipes
by the user - Having somewhat greyish results for
- Shift towards the achromatic axis in proportion
to - Causing a chroma reduction for all points ((c)
recipe)
Figure 12. The result for sharpening within the
gamut (a) for different values of , with
.
18Conclusions
- Proposed method
- A generic method that allows grey image
processing for lightness processing on color
images - Without affecting color rendering
- Staying within the gamut of the apparatus in
question - Results
- Sharpening improvement
- The chroma change in proportion to the lightness
- Improvement for the most images
- Contrast improvement
- The chroma reduction in proportion to the
lightness difference - Mapping towards black and white recipe
- The chroma increase in proportion to the
lightness difference - Mapping away from black and white recipe