Detection and reduction of impulse noise using neuro-fuzzy system and dilation rule

Restoration of digital images contaminated by impulse noise is still a challenging job for researchers. A novel filter based on neuro-fuzzy system and dilation rule has been proposed, which works well in presence of high density noise. Adaptive Neuro-Fuzzy Inference System (ANFIS) has been used to detect the impulse noise, and dilation rule has been used to remove the detected corrupted pixel. ANFIS helps to preserve the fine details of the image and a little modified dilation rule gives the estimated value for pixel replacement. Experimental results show the effectiveness of the proposed restoration method by qualitative and quantitative analysis.