An Adaptive Weighted Min-Mid-Max Value Based Filter for Eliminating High Density Impulsive Noise

Abstract

This paper presents a novel algorithm to filter impulsive noise at very high noise density ( $$\ge 85\%$$ ). The proposed algorithm initially makes an accurate decision and selects a window which has sufficient information for denoising. Within the selected window, the proposed algorithm computes maximum, minimum, middle values along with their weights to restore noisy pixel. The performance of proposed filter is evaluated on natural and medical images with varying noise density. The proposed filter showed tremendous performance at high noise densities in terms of quantitative metrics and visual representation. Even at noise densities as high as 97% and 99%, the proposed filter is able to retrieve the details of the image. The proposed filter on an average improves the peak signal to noise ratio value by 10% in medical images over the existing.