Research on change detection of high resolution remote sensing image based on U-type neural network

Abstract

The change detection of high-resolution remote sensing images has always been a research difficulty and hot spot, which has a great application demand in land use change and ecological environment monitoring. U-type neural network can train a better model with less samples, and has good experimental effect in medical image segmentation, but it is rarely used in remote sensing image change detection research. In this paper, the U-type neural network is used to detect the change spots in GF-1 image of Yuzhou City, Henan Province, and compared with FCN, SegNet and Siamese-CNN. Experimental results show that the F1 score of U-type neural network model are 0.699 and 0.66 respectively, which are better than the other three methods, and the omission is lower, which is closer to the label map. Therefore, it is feasible to use U-type neural network for change detection of high-resolution remote sensing images, and it can have high detection accuracy.