Fuzzy vector partition filtering technique for color image restoration

A fuzzy vector partition filter is proposed for real color image restoration applications. The filter utilizes the fuzzy ranking technique in the vector partition filtering framework to address a variety of noise corruptions in natural color images. The information of the local sample spread is tracked by the fuzzy ranking through its adaptive fuzzy sample similarity function. The non-stationary statistics of the color image is overcome by partitioning the image into a finite set of underlying local image structures and by applying an optimized nonlinear filter to restore each of the partitions accordingly. The proposed filter has been evaluated using various simulated and real-life noisy images. The results achieved by the filter are comparable to the prior-art methods specially designed for the underlying noise corruption, in terms of standard objective measurements and the subjective picture quality.