UNDERWATER IMAGE ENHANCEMENT USING DWT BASED MEAN IMAGE FUSION RULE

Abstract

Due to the physical properties of water and its environment, underwater image processing has received a considerable attention since last decade. Underwater activities in discovering and recognizing objects have resulted in new challenges. Consequence from these activities, significant problems have raised due to the light absorption and diffusion effects. As light travels in water, a rapid exponential loss of light intensity occurs depending on the color spectrum wavelength. In this work, Transform based method has been used to preserve the contrast and color of the images. At first, Color and contrast corrected images has been evaluated using HSV color space and CLAHE method of adaptive histogram equalization. Then these two images has been fused together using different fusion rules at two level decomposition using DWT. Different fusion rules has been explored in which mean od approximation and detailed coefficient found effective than the rest. It gives high PSNR values along with improved contrast and entropy of the image. Experimental results have been compared between different methods of fusion by PSNR, MSE, Average gradient, entropy and contrast parameters. Proposed method has been tested on number of images and results found at various stages of the algorithm have been displayed. In future, similar work can be explored further by taking different transform based decompositions as only DWT has been tested in this work. Also Different wavelet families have been explored during work and ‘Bior6.8’ found to be effective one than the other.