Research on salt dome recognition algorithm based on reverse attention mechanism

In view of the strong subjectivity of traditional salt dome recognition methods and the poor effect of existing deep learning algorithms on salt dome edge recognition, this paper proposes a salt dome recognition algorithm based on reverse attention mechanism, which uses u-net model as the backbone network, adds reverse attention module at the jump connection to extract edge structure information, and finally uses feature splicing to fuse feature information to improve the segmentation performance of network model. Experimental results show that the network achieves good results in salt dome segmentation, and effectively improves the problem of unclear edge segmentation.