Adaptive Vector Filters for Impulse Noise Removal in Color Image

In this paper, new filters based on noise detection are proposed for impulse noise removal in color images. In the new method, impulse noise is properly detected, and the detected noisy pixels are filtered with adaptive selection of filter window, while the noise-free pixels are kept without any changes. Experimental results show that the proposed filters remove impulse noise and preserve edges and details very well, which perform better than the classical vector median filter (VMF), direction-distance filter (DDF), and directional-magnitude vector filter (DMVF).