A switching vector median filter based on the CIELAB color space for color image restoration

Vector median filter is highly effective in removing impulsive noise from color images. However, it fails to distinguish thin lines and boundaries from impulsive noise, and usually filters them out because it interprets these fine details as some noise. This paper introduces a new solution for impulsive noise detection in color images. The proposed solution is a switching vector filter which analyzes the color difference of two pixels in the CIELAB color space using four directional operators. Based on the result of this analysis, the impulse detection module can identify noise pixels, which are replaced in the filtering module with some robust estimate. The extensive experimental results show that the proposed solution outperforms many of the existing vector filters in terms of the filtering performance. In particular, the proposed approach can effectively preserve the thin lines, fine details, and image edges.