Video restoration method and device, and medium

Abstract

The invention discloses a video restoration method and a device, and a medium. The method comprises the following steps: obtaining a first training data set; extracting a first degradation representation of each first low-resolution video in the first training data set by using a trained feature extraction network; inputting the first degradation representation and the first low-resolution video into a pre-constructed super-resolution network to obtain a predicted high-resolution video; determining a first error between the predicted high-resolution video and the high-resolution video, and updating the super-resolution network based on the first error until the super-resolution network converges to obtain a trained super-resolution network; and when a to-be-repaired video is obtained, extracting a second degradation representation of the to-be-repaired video by using the trained feature extraction network, and inputting the second degradation representation and the to-be-repaired video into the trained super-resolution network to obtain a repaired high-resolution video. In this way, the video can be repaired according to the noise brought by different compression coding modes, and the video repairing effect is improved.

Description

technical field [0001] The present application relates to the technical field of image processing, and in particular to a video restoration method, device, and medium. Background technique [0002] The existing video quality improvement mainly uses algorithms such as noise reduction, deblurring, and super-resolution to improve the image quality and resolution of video files. Among them, the noise reduction and deblurring algorithms mainly use low-pass filtering and other methods to filter and reduce noise and Gaussian sharpening on the picture. The super-resolution method mainly aligns multiple video frames by calculating the optical flow, and extracts features by using the redundant information of adjacent video frames. The blurred places in the current frame can be borrowed from adjacent frames and used for Restore high-resolution images. [0003] In the above-mentioned prior art, various distortions that may exist in real video data due to compression coding methods are...

AI Technical Summary

Problems solved by technology

[0003] In the above-mentioned prior art, various distortions that may exist in real video data due to compression coding methods are ignored, for example, MPEG-2, HEVC compression noise, etc., resulting in poor video repair effect

Images