Binarized switching median filter with pre-segmentation for impulse noise removal

Abstract

To restore image corrupted by salt-pepper impulse noise, a new switching median filter, pre-segmentation binarized switching median (PBSM) filter, is presented, which is based on adaptive thresholds decomposition. The PBSM filter consists of two stages. The binarized switching median (BSM) filter is developed to classify each pixel to be uncorrupted pixel or impulse noise so as to decide whether to deal with standard median (SM) filter or not. Segmentation uses Otsu's segmentation method to adaptively separate image for PBSM filter becoming effective. Experimental results show that this proposed filter provides better performance than standard median filter.