Image rain removal method and device based on PReNet progressive network

The invention discloses an image rain removal method and device based on a PReNet progressive network. The method comprises the following steps: S1, obtaining a truly rendered image to be subjected to rain removal and corresponding precipitation information; s2, gradually extracting feature information of the image in a mode of gradually increasing the resolution; s3, a progressive residual block is adopted as a core module, and feature information in the image is extracted; s4, fusing feature information of different resolutions by using a feature fusion network; s5, rain removal is completed; s6, outputting the rain-removed images of the stages 1-N; and S7, training the model by adopting a polynomial regression algorithm. According to the image rain removal method and device based on the PReNet progressive network, the discrimination of traffic signs in rainy days is improved, and the speed and accuracy of a whole traffic target detection and recognition system are improved. 本发明公开了一种基于PReNet渐进式网络的图像去雨方法及装置，