Improvement of underwater image contrast enhancement technique based on histogram modification
Loading...
Date
2015-03-01
Authors
Ahmad Shahrizan Abdul Ghani
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Contrast degradation is one of the problems of underwater image that resulted from the light
attenuation. Low contrast contributes towards the less usable image where less information
could be extracted from the image. The objects seen in the image are unclear. In addition,
light absorption phenomenon causes the underwater image to be dominant by the blue-green
illumination, resulting in misinterpretation of objects color. Therefore, to reduce the
aforementioned problems of underwater image and increases underwater image contrast, three
techniques of improving underwater image contrast are proposed in this study, namely
integrated color model with Rayleigh distribution (ICM-RD), Rayleigh-stretching and
averaging image planes (RSAIP), and dual-images Rayleigh-stretched contrast limited
adaptive histogram specification (DIRS-CLAHS). ICM-RD improves the underwater image
contrast by integrating the Rayleigh distribution in the limited stretching process. The
correction of image color through Hue-Saturation-Value (HSV) color model further improves
the overall image color. On the other hand, RSAIP method solves the limitation of stretching
process that faced by ICM-RD method. The RSAIP method provides an alternative stretching
technique, where the histogram of the original image is divided into two independent regions
and stretched independently to occupy the limited dynamic intensity range. The dividing and
stretching processes produce two different intensity images. These images are then combined
by means of average value and applied with color correction technique to produce final
resultant image. The third proposed method, DIRS-CLAHS method is designed to improve
the capability of the RSAIP method in enhancing image contrast by integrating global and
local contrast correction. DIRS-CLAHS is first applied with global contrast correction which
is introduced in the RSAIP method. Local contrast correction is then applied by dividing the
image into smaller tiles. Finally, the method is applied with a new color correction process
which is a modification of color correction process introduced in RSAIP and ICM-RD
methods. All proposed techniques, principally outperform the state-of-the-art methods,
qualitative and quantitatively. Out of the three proposed methods, DIRS-CLAHS method, is
the best method and demonstrates a significant enhancement in improving the underwater
image contrast and its color. Quantitatively, in comparison with six state-of-the-art methods
for 300 samples of underwater images, the proposed DIRS-CLAHS produces the highest
average entropy of 7.624 and the lowest average MSE value of 646.32. In addition, in terms
of measure of enhancement (EME) and measure of enhancement by entropy (EMEE), DIRSCLAHS
produces the highest average values which are 27.096 and 9.670, respectively.