29-10-2012, 01:42 PM
Efficient Generalized Colored image Enhancement
Efficient Generalized.pdf (Size: 202.51 KB / Downloads: 22)
Abstract
Hue preservation is an essential requirement for color
image enhancement, but preserving hue introduces the gamut
problem. A fast technique is proposed which can be applied to
generalize any linear or non-linear gray-scale contrast
enhancement function to the color domain. The technique
provides hue preserved, gamut problem free color contrast
enhancement in accordance with the gray-scale contrast
enhancing function it generalizes. Results for the proposed color
image enhancement technique are better than the currently
available techniques.
Introduction
Image enhancement is used to improve the quality of
an image for visual perception of human beings. The main
techniques for image enhancement such as contrast stretching,
slicing, histogram equalization etc, for grey scale images are
discussed in many books. The generalization of these
techniques to color images is not straight forward. Unlike grey
scale images, there are some factors in color images like hue
which need to be properly taken care of for enhancement.
Hue, saturation and intensity are the attributes of
color [1]. If a saturated color is diluted by being mixed with
other colors or with white light, its richness or saturation is
decreased [2]. For the purpose of enhancing a color image, it
is to be seen that hue should not change for any pixel. If hue is
changed then the color gets changed, thereby distorting the
image. For image enhancement, one needs to improve the
visual quality of an image without distorting it.
Gray Scale Contrast Enhancement
The gray scale contrast enhancement techniques involve
scaling and shifting operations; the net result of these
operations on an image is that all its pixels with gray values
above a certain reference point, w.r.t. that particular image, are
pushed to a higher value while all the pixels with gray level
below that point are pushed to lower gray values. Generally,
the farther a gray value from that point, the more outward
(w.r.t. that point) it is pushed. Notice that if the above
mentioned reference point occurs at either of the upper or
lower boundary, all gray levels are pushed in one direction
only, in which case, contrast is enhanced by stretching of the
gray levels in the gray domain.
The Drawback of Intensity Based
Generalization
Having discussed the basic principles of gray scale
contrast enhancement, one can observe the drawback in the
intensity based generalizations to colored images. The hue
preserving generalization implies that each of the R, G and B
values are pushed upwards/downwards by the same factor,
hence the generalization of the grey-scale function to the color
domain has to be a linear transformation. The drawback is that
when intensity ‘I’ is pushed upward (higher from its initial
value) or downward, depending upon the value of the
reference point mentioned above, it is possible for either of the
(normalized) R, G or B value to be on the other side of the
reference point than that of the (normalized) intensity value. In
such cases, those R, G or B values would be pushed in the
opposite direction to that in which the gray-scale function
would push the equivalent grey values. This results in a
decrease (instead of enhancement) in the contrast of the R, G
or B values of such pixels from the other pixels.
The Gamut Problem
Another problem encountered in the intensity based
generalizations is the gamut problem, in which the factor
calculated from the intensity value, by which each of the R,G
and B values is scaled, scales a very high R,G or B value out
of its allowed domain. None of the above referenced
techniques solve this problem except [12], [14] and [16]. To
take care of the R, G, and B values exceeding the bounds,
weeks et all [12] suggest normalization of each component
using (255/max(RGB)), This technique, however, makes the
image darker. Young et al have developed clipping techniques
[14] in LHS and YIQ spaces to take care of the gamut
problem. Clipping is performed after the enhancement.
Clipping however distorts the hue of the image which is not
desirable. Naik et al have also avoided the gamut problem in
their algorithms [16] for generalization of linear and nonlinear
gray contrast enhancement functions.
Conclusion
From the above discussion and experimental results, it
can be concluded that simple techniques developed using the
basic principles related to a problems domain can surpass
highly complicated and mathematically elegant techniques
with respect to the quality of the output produced. The simple
efficient technique proposed in this article produces better
results than the best of a number of earlier techniques implied
for the solution of the same problem. And these better results
are produced in spite of the fact that this proposed algorithm is
much faster than the others addressing the same problem.
Hence, the proposed algorithm of this article is an efficient
and reliable choice for hue preserving, gamut problem free
contrast enhancement of colored images.