Optoelectronic implementation of a multifunction cellular neural network

Image feature extraction by dynamic neural filtering and phase-only joint transform correlation

“…Furthermore, the possibility of optical implementation is the other main advantage of the proposed technique which can be incorporated with various optical image processing or optical pattern recognition techniques just in the same optical system. Compared to the previously reported optical neural network-based image feature extraction techniques [10,[13][14][15], dynamic neural filtering technique presents high-speed processing ability due to the utilization of minimum number of interconnections terms and minimum optical operations. For instance, Cellular neural network achieves same image feature extractions by two weight masks, while the proposed technique can be carried out just by one filtering mask.…”

“…(13) and (14). When the learning algorithm converges to minimum error, last updated filter and bias coefficients are saved for the test phase.…”