Neural networks for image processing: New edge detection algorithm
Giuseppe Grassi, Pietro Vecchio, Donato Cafagna, Eugenio Di Sciascio, Luigi Alfredo Grieco · 2007
Neural networks can be very useful for image processing applications. This paper exploits the cellular neural network (CNN) paradigm to develop a new edge detection algorithm. The approach makes use of rigorous model of the image contours, and takes into account some electrical restrictions of existing CNN-based hardware implementations. Four benchmark video sequences are analyzed, that is, Car-phone, Miss America, Stefan, and Foreman. The analysis shows that the proposed algorithm yields accurate results, better than the ones achievable by other CNN-based methods. Finally, comparisons with standard edge detection techniques (i.e., LoG edge detector and Canny algorithm) further confirm the capability of the developed approach.