Two-stage quaternion switching vector filter for color impulse noise removal

Abstract

In this paper, a new approach to impulse noise removal in color images is presented. The proposed solution is a quaternion switching vector filter in which the impulse detection consists of two stages. First, by using quaternion representation, an effective color distance measure method is developed. Then, based on the new color distance measure, the proposed filter utilizes the directional samples along four directions to classify image pixels into possible noisy and noise-free ones. For possible noisy pixels, the concept of peer group is modified and extended to the directional samples to further detect whether they are corrupted by impulse noise or not. Finally, a weighted vector median filter is performed only on the pixels that are identified as noisy by the second stage. The experimental comparisons exhibit the validity of the proposed solution by showing significant performance improvements over other well-known color image filtering methods.