Effective solution for digital image edge detectionbased on artificial neural network

In this paper, an artificial neural network system isused to identify the edges of digital images. In the proposedmethod, for each pair of image in the database (original imageand real edge), a table containing edge and non-edge records isextracted. Then using the k-mean clustering algorithm, theaverage number of non-edge records should be reduced. Byreducing the volume of records, the information in the tables iscombined together and the training table for the neural networkis created. To test the neural network by grabbing a gray image,it is divided block wise and each block is delivered to the neuralnetwork as an input, the neural network gives the output valueas an edge or non-edge pixel. To analyze the results, differentvarious metrics have been used to evaluate the result of theproposed method compared to other methods. The results of theimplementation show that the proposed algorithm performsbetter than comparable methods

Keywords:

edge detection , artificial neural network , k-mean clusteringalgorithm , and image blocking.