A new adaptive filtering method for removing salt and pepper noise based on multilayered PCNN

Image de-noising is an important image preprocessing technique, especially how to restore the image corrupted by high density salt and pepper noise is a hotspot of current research. This paper presents an adaptive de-noising method by using the multilayered Pulse Coupled Neural Network (PCNN). By analyzing the firing characteristics of the improved PCNN, a new approach is proposed for setting adaptively the parameters of PCNN based on the elimination of “Mathematics Firing”. Through this method, the salt and pepper noise can be located adaptively by the PCNN, and then we propose an improved median filtering method which only uses uncontaminated pixels to determine the median. Finally, by repeating both the noise location and removal of the noise, the varying density salt and pepper noise in the image can be removed gradually. Simulation experiments show that the proposed method performs better compared with the state-of-the-art method for the varying density salt and pepper noise, especially for high density salt and pepper noise, the proposed method performs well on the aspect of removal of salt and pepper noise and protecting image details.