High density impulse noise removal using BDND filtering algorithm

Abstract

Switching median filters outperform standard median filters in the removal of impulse noise. This is because, it considers only the noisy pixels and performs filtering operation on that pixels without considering noise-free pixels. The Boundary Discriminative Noise Detection (BDND) filter is proven to operate effectively under different impulse noise models. It initially classifies pixels into three groups as (a) low intensity impulse noise (b) high intensity impulse noise (c) uncorrupted pixels. Then noise detection and filtering steps are performed. Pixel misclassification is the main drawback of BDND filtering algorithm. So we modify the filtering step of this algorithm and named it as modified boundary discriminative noise detection (MBDND). The two modifications incorporated are as follows: (1) Expansion of filtering window. (2) Incorporating spatial and intensity information. By introducing these modifications into the algorithm, it is found that there is increase in the performance and the quality of image has improved. Results are compared with other median filters like Center Weighted Median Filter (CWMF), Progressive Switching Median Filter (PSMF), Adaptive Threshold Median Filter (ATMF) and it is found that MBDND performs well even at high noise density (90%).