Modified decision based unsymmetric adaptive neighborhood trimmed mean filter for removal of very high density salt and pepper noise

Abstract

In this paper, a novel and efficient algorithm, Modified Decision Based Unsymmetric Adaptive Neighborhood Trimmed Mean Filter, for removal of very high density salt and pepper noise (SPN) is proposed. The proposed technique comprises of two phases. The first phase involves the fusion of decision based partial trimmed global mean filter, decision based unsymmetric trimmed modified Winsorized mean filter and decision based adaptive neighborhood median filter which takes the benefits of partial trimmed global mean, unsymmetric trimmed modified Winsorized mean and neighborhood pixel technique. The second phase is applied to wipe out the left over noisy pixels by replacing the processing pixel with the Winsorized mean of the first ordered neighborhood pixels. The proposed algorithm is examined upto 99% levels of salt and pepper noise for grey scale and color bitmap images and it gives better Peak Signal to Noise Ratio (PSNR), Image Enhancement Factor (IEF) and Structural Similarity Index (SSIM) values for high noise densities.