Road extraction from high-resolution remote sensing images based on HRNet
In response to the problems of low accuracy and weak robustness in traditional road extraction from high-resolution remote sensing images,a high-resolution net(HRNet)based approach is proposed to achieve road segmentation in high-resolution remote sensing images.This article improves HRNet by concatenating the output of HRNet subnets of the same resolution with the output layer results and inputting the above into non-local block.The two loss functions,Cross-entropy Loss and Dice Loss,are used to solve the problem of imbalanced road dataset samples.The experimental results show that the improved HRNet performs better in road extraction on the publicly available CHN6-CUG road dataset compared to other methods,achieving 97.65％,84.91％,and 97.25％respectively in recall,mean intersection over union(MIoU),and F1 score.
万方数据
维普
