Noise Density Range Sensitive Mean-Median Filter for Impulse Noise Removal

Abstract

A new noise density range sensitive algorithm for the restoration of images that are corrupted by impulse noise is proposed. The proposed algorithm replaces the noisy pixel by mean, median or pre-processed values based on noise density of the image. The proposed filter uses a unique approach for recovering images corrupted with very high noise densities (over 85%). It also provides significantly better image quality for different noise densities (10–90%). Simulation results show that the proposed filter outperforms in comparison with the other nonlinear filters. At very high noise densities, the proposed filter provides better visual representation with 6.5% average improvement in peak signal-to-noise ratio value when compared to state-of-the-art filters.