An Application of CP Neural Network Based on Rough Set in Image Edge Detection

Abstract

Based on the rough set theory, a counter propagation neural network algorithm for edge detection is presented in this paper. Firstly, a definition of rough membership function, which is used to modify the weigh values in the nomal counter propagation neural network, is proposed after introducing the rough set. Experiments show that the approach has achieved good results in improving the accuracy of detection. And this algorithm can also overcome effectively the problem of simply cluster in the nomal counter propagation neural network.