Dual-channel single image fine rain removal method

Abstract

The invention discloses a dual-channel single image fine rain removal method, and provides a new dual-channel residual dense block module (dual-channel hybrid block for short) for accurately obtaininga negative residual rain streak feature map, which comprises a residual path and a dense path. The residual path is used for reusing a common rain streak feature map from a front layer of the deep convolutional network, and the dense path can be used for exploring a new rain streak feature map. On the basis of a dual-channel mixing block, a cascaded dual-channel mixing block based on the dual-channel mixing block is constructed for rain streak feature extraction. In order to connect the characteristics of different scales, the method also adopts the idea of multi-stream branches, and data link channels are arranged between the multi-stream branches and between streams based on dual-channel residual dense blocks for rain streak characteristic information sharing. After a rain streak feature map of multiple sensing domains is obtained and a rough negative residual rain streak feature map is obtained through a convolutional neural network, a final accurate rain-removed image can be finally obtained through fine tuning.

Description

technical field [0001] The invention relates to the field of computer vision image processing methods, in particular to a method for finely removing rain from a single image with two channels. Background technique [0002] Removing rain streaks from images is an important and challenging topic in numerous computer vision and data mining tasks, such as autonomous driving, drone-based video surveillance, and real-time object recognition under severe weather conditions. Rain is the most common type of weather that can reduce image quality. Due to the interference of rain streaks, some advanced tasks such as object detection, image recognition, and saliency detection may be affected. Therefore, novel and effective methods are developed to automatically process images in images. The rain streaks are very important. [0003] The rainy image mainly contains rain streaks and raindrops, which form a rain mask before the real image. For torrential rains, they may create a hazy atmos...

AI Technical Summary

Problems solved by technology

However, due to the complex ill-conditioned binary problem in the mathematical model of the rain streak removal problem, the above method cannot extract accurate rain streaks from the rainy image, especially during the training process, if the rain streak direction in the test image is not properly considered and scale, it is often prone to excessive rain removal (resulting in blurred images) or a small amount of rain removal (resulting in a large number of remaining rain streaks)

The main reason is that the rain pattern learning and expression ability of these methods are insufficient, especially under the condition of heavy rainfall (that is, the rain pattern is large and dense), and the mapping relationship between the rainy image and the non-rainy image cannot be accurately learned.

Images