A single image rain removal method based on multi-scale aggregation features

Abstract

The invention belongs to the field of computer vision and relates to a method for removing rain from a single image based on multi-scale aggregation features. The present invention is based on a multi-scale feature aggregation densely connected convolutional network framework, which consists of an encoding-decoding network, and each encoding network corresponds to a decoding network; the encoding network performs maximum pooling through dimensionality reduction and downsampling of the maximum pooling layer. The index position of the maximum pooling is recorded during the process of pooling, and the pooling index guides the upsampling recovery process of the corresponding decoding network. Among them, the encoding network and the decoding network are the same in the convolutional layer, both of which are feature aggregation densely connected convolution modules, only the maximum pooling and its corresponding upsampling process are different. The invention can effectively remove the rain streaks with different densities, and at the same time well preserve the details of the image.

Description

technical field [0001] The invention belongs to the field of computer vision and relates to a method for removing rain from a single image based on multi-scale aggregation features. Background technique [0002] In rainy days, the impact of rain marks on images and videos is often undesirable, and rain marks will seriously affect the performance of many outdoor computer vision applications, such as surveillance systems and automatic driving systems. In the field of security and outdoor photography, these camera and camera equipment are easily affected by severe weather such as severe rainstorms when working outdoors. However, in general, the design of the camera system for collecting images outdoors in practical applications does not take into account the outdoor weather. influences. When encountering weather conditions such as heavy fog, heavy rain, and heavy snow outdoors, the visibility of people's sight and the clarity of camera equipment will be greatly affected. The p...

AI Technical Summary

Problems solved by technology

However, some of these algorithms have incomplete description of the characteristics of the rain line, resulting in unclean rain removal. The results of some algorithms lead to excessive smoothing of the image, and the restoration of the image background is not complete enough.

Images