Performance enhancement of image impulse noise filters by image rotation and fuzzy processing

A novel filtering approach for improving the performances of recursive impulse noise filters is proposed. The method is based on the fact that the output of a recursive image filter varies depending on the orientation of the input image. Hence, the proposed method generates four input images by rotating the noisy input image at integer multiples of 90°90°, obtains four restored images by processing these images by the noise filter, and then computes the enhanced output image from these filtered images by using a fuzzy system. Internal parameters of the fuzzy system are determined by training by using simple artificial training images. The validity of the proposed method is demonstrated by extensive filtering experiments, which are conducted for four different impulse noise filters, four different test images and three different noise densities. The results show that the proposed method can efficiently be used for improving the performance of any recursive impulse noise filter to obtain significantly enhanced output images.