A Method of Image Deraining Based on Cross-Domain Collaborative Learning

Abstract

The invention discloses an image deraining method based on cross-domain collaborative learning, constructs a dual-branch image deraining network based on a multi-scale attention residual module, and reduces rain in different synthesis fields through a cross-domain collaborative learning strategy oriented to the synthesis field The impact of the difference in the distribution of grain information on the deraining effect of the dual-branch image deraining model, through the real domain-oriented cross-domain learning strategy, reduces the influence of the difference in the distribution of rain grain information in the real domain and the synthetic domain on the deraining performance of the dual-branch image deraining model . The invention improves the learning ability of the image deraining model for the rain pattern information of samples in different fields, reduces the influence of the difference in the distribution of rain pattern information in different fields on the deraining performance of the image deraining model, and enhances the robustness and performance of the image deraining model. Generalization.

Description

technical field [0001] The invention relates to the field of image restoration and enhancement, in particular to an image rain removal method based on cross-domain collaborative learning. Background technique [0002] Photos taken under rainy weather conditions are often of low quality, and these image quality degradations due to rain streaks will affect a range of computer vision tasks, e.g., object detection, image recognition, etc. Therefore, designing effective image deraining algorithms is crucial to the practical application of computer vision algorithms. [0003] At present, deep learning has achieved excellent performance in image deraining tasks. For example, Zhang et al. proposed a single image de-raining model based on conditional generative adversarial network, and used the perceptual loss function to further improve the de-raining effect (He Zhang, Vishwanath Sindagi, and Vishal M. Patel, "Image de-raining using a conditional generative adversarial network”, I...

AI Technical Summary

Problems solved by technology

[0005] The purpose of the present invention is to provide an image deraining method based on cross-domain collaborative learning, and at the same time solve the problem of the rain pattern information distribution difference between different synthetic domains and between the synthetic domain and the real domain for the deraining performance of the image deraining model Influence, improve the robustness and generalization ability of the image deraining model

Images