Vector median filters for processing of color images in various color spaces

The paper presents a comparative study of performance of 2D vector median filters which calculate in various color coordinates the distances between the vectors corresponding to pixels of color images. The paper deals with the problem of the best choice of the color space for vector median calculations. The problem is considered for the task of restoration of color images corrupted with impulse noise. The comparison has been made experimentally, using 4 test images and 3 types of impulse noise. The results have been evaluated by consideration of quadratic errors in the Munsell space as well as by comparing their subjective quality.