Hybridization of Algorithm for Restoration of Impulse Noise Image

Abstract

In today's era, mainly communication is done through visual communication. Almost all information is transmitted in the form of digital image or video. But after transmission, the obtained information is often corrupted with noise. At a high noise density, detail information of the image is hidden by noise. Hence, we have to recover the original image by removing noise of the image without loss of data. Here, we proposed two hybridization methods, to yield better restoration of impulse noise images. The first proposed algorithm has hybridization of Decision Based algorithm along with 2D discrete wavelet transform. In second, hybridization of Decision based algorithm with Adaptive Wiener Filter. Experimental results in Figs. 3–6 shows that proposed algorithm 1 outperform in terms of visual quality till 80% of noise density. Proposed algorithm 2 also performs excellent in term of visual quality, but within noise density ranges from 70% to 90%. Even at 95% noise density, proposed algorithm 2 gives an improvement in PSNR value from 5.29 dB to 18.98 dB. Mean Absolute Error, Mean Square Error, Peak Signal to Noise Ratio and Image Enhancement Factor are calculated and the comparative study is made between Standard Median Filter, Cascaded decision based algorithm proposed in10, and proposed algorithm.