Rain removing method based on image restoration technology

Abstract

The invention discloses a rain removal method based on an image restoration technology. The method comprises the following steps: constructing a coarse rain removal network composed of three feature extraction modules and a residual image prediction module; constructing an image fine restoration network composed of a threshold judgment module and an image restoration module; training the coarse rain removal network and the image fine restoration network based on the rain image database distribution, and combining the trained coarse rain removal network and image fine restoration network into arain removal model based on the image restoration technology; and processing the input image based on the rain removal model. According to the method, the utilization of the feature information of the residual image can be accurately and effectively realized, the rain removal precision can be obviously improved, and the actual scene application of the rain removal algorithm is further promoted.

Description

technical field [0001] The invention relates to the fields of deep learning and computer vision, in particular to a rain removal method based on image restoration technology. Background technique [0002] Rainy day is the most common severe weather, in which rain lines in various directions, densities, and sizes will block and blur the content of the captured image to varying degrees, directly leading to poor subjective visual effects. In addition, poor-quality images can significantly degrade the performance of image-based computer vision tasks such as object classification, detection, and segmentation. At present, the rain removal methods can be roughly divided into two categories: one is the traditional method based on dictionary learning, sparse coding, etc.; the other is the deep learning method based on multi-layer nonlinear convolution. Compared with the deep learning method, the rain removal image obtained by the traditional method is smooth and blurred, resulting i...

AI Technical Summary

Problems solved by technology

However, these methods do not accurately use the feature information of the residual image to improve the rain removal image.

Images