Image Restoration Using Fuzzy Impulse Noise Detection and Adaptive Median Filter

In this paper we propose a new algorithm to reduce the impulse noises in images. It contains two separated steps: impulse noise detect step and filter step. In impulse noise detect step we use the fuzzy gradient method with new fuzzy rules to distinguish the noise pixels and edge pixels; and filter step employs a novel adaptive median filter whose size is controlled by the number of interesting pixels in the filter window and its dynamic number threshold which will be defined in this paper. Experimental results show it is effective in image impulse noise removing while preserving the image details.