Image Noise Removal And Edge Detection Based On PCNN

This thesis firstly gives a detailed analysis of the mechanism of Pulse-coupled Neural Networks (Simplified as PCNN), and then emphasizes its applications on image noise removal and edge detection. PCNN is a new type of network and is called the third generation artificial neural network. It is a simplified model built through the simulation of the outbursts of synchronous pulses in the visual layer of a cat's cerebra. At present, much attention is paid to the research on the mechanism of PCNN and its applications on image processing, automatic target recognition, combination and optimization, attention and artificial life abroad. Yet less work has been done at home. In this paper, the main work includes: (1) Through the detailed analysis of the mechanism of PCNN, we point out that the behaviors of PCNN actually means the reorganization of input information, and finally two methods (including method for salt & pepper noise and method for Gaussian noise) for image noise removal based on PCNN are presented. (2) Given the features of image edge, a new function named lateral inhibition is added to PCNN, and a new algorithm for image edge detection based on PCNN is also produced. The whole topic is full of new ideas, and shows its great advantages over other methods.