Dual-channel convolutional neural network for change detection of multitemporal SAR images

This paper presents a new dual-channel convolutional neural network (CNN) which is designed to SAR image change detection to acquire higher detection accuracy and lower misclassification rate. This network model contains two parallel CNN structures, which can extract features from two multitemporal SAR images. Experimental evaluation on simulated datasets and real SAR images from different satellites shows a satisfying performance of the proposed model. It is the second-place winner when detecting the simulated images with Gamma noise, but it wins the top place when detecting the real flood images.