Remote sensing image change detection method combining feature fusion and enhancement difference

A remote sensing image change detection method combining feature fusion and enhancement difference comprises the following steps: preprocessing a two-stage remote sensing image, inputting the preprocessed two-stage remote sensing image into a model, extracting the features of the two-stage remote sensing image through a down-sampling channel, and calculating an enhancement difference feature map; cascading the highest layer features of the two channels and the difference features thereof as up-sampling input, establishing an auxiliary up-sampling channel in the up-sampling process to fuse the features to a main up-sampling channel, and cascading the enhanced difference features of each scale and the shallow layer features in down-sampling to corresponding up-sampling layers; and performing deconvolution and Sigmoid operation on the output of the last layer of up-sampling to obtain an image change probability graph, calculating loss according to a real image change label by using a mixed loss function based on