Optronic Convolutional Neural Network for SAR Target Recognition

Abstract

Using deep convolutional neural networks to achieve automatic target recognition (ATR) is effective but it will bring heavy computation burden. Here we propose an optronic convolutional neural network (OPCNN) to realize ATR in optics. By using OPCNN, the computation cost is dramatically reduced and good performance of recognition accuracy is obtained simultaneously. Experimental results on Moving and Stationary Target Acquisition and Recognition dataset demonstrate the feasibility of our proposed OPCNN architecture.