Removal of High Density Salt and Pepper Noise from Color Images through Variable Window Size

In this paper we propose a new efficient algorithm for restoration of both color and gray scale images affected by Impulse noise. The algorithm framed works adaptively at different noise levels. Based on noise density, pixel is healed by calculating mean of surrounding healthy pixels in a considered window size and taking the center pixel as the kernel of the window. The proposed algorithm restores pixels which are badly affected by fixed valued impulse noise (0,255)(In 8-bit encoding). The algorithm processes the whole image in a single pass and heals the affected kernel in the considered window by measuring successive amplitudes of pixel. The proposed algorithm also aims in preserving edges of images and thus distortions measured in edges are sufficiently less than other proposed algorithms. Computer simulations verify substantial improvement over other methods in terms of Peak-Signal-to-Noise Ratio (PSNR) and Visual quality.