Edge Detection of Image Based on Improved BP Network

A kind of edge detection method is presented,which trains the BP neural network by an improved BP algorithm. Disturbed by noise,edge of image can not be detected finely by normal method. Therefore the essential differences between noise and edge are fully considered while selecting the image features and forming feature vector. The experiment use sample image to train BP neural network,then this trained network is used to detect images' edge. The experimental results show that this method is very effective.