IMAGE EDGE DETECTION USING ARTIFICIAL NEURAL NETWORK

A.V. Naumenko, V.V. Lukin

A method of joint analysis of several local edge detector outputs using an artificial neural network is proposed. The efficiency of applying different edge detectors in the neural network edge detector is considered. It is established that the usage of only the most informative local parameters can improve characteristics of the neural network edge detector. A procedure of neural network learning is described and comparison of neural network detector efficiency to several known detectors is carried out. The tasks of further research are formulated.

Keywords: edge detection, neural networks, joint analysis of local parameters