An Intelligent Impulse Noise Detection Method by Adaptive Subband-Based Multi-State Median Filtering

This paper presents a novel random-value impulse noise reduction method by adaptive subband-based multi-state median (ASBMSM) filtering. Most reported noise filters can effectively reduce impulse noise distributed over the low-frequency area in an image but hardly filter these noises distributed over the high- frequency area. To overcome the above problem, we introduce an adaptive subband-based filtering strategy in which an image is divided into low-frequency blocks and high-frequency blocks of size 8*8 by using PSNR (peak signal-to-noise-ratio) checking and then the appropriate mask for each type of block is employed. In addition, to substantially restore the original information, another filtering process is required if the PSNR value of the previously filtered image is lower than a threshold. Experimental results show that the proposed impulse noise filter manifests the superiority in PSNR performance to other switching-based median filters when the corruption ratio is below 30%.