Video processing method and device based on super-resolution network, and storage medium

Abstract

The invention discloses a video processing method, which comprises the following steps of: inputting each low-resolution image frame of a low-resolution video into a trained super-resolution network for processing to obtain a super-resolution image frame; wherein the shallow feature of each low-resolution image frame is extracted through the shallow feature extraction convolution layer; depth features of the shallow layer features are extracted through a residual network structure RIR module; extracting depth features of the shallow layer features through a preset residual group RG; in the RG, depth feature extraction is carried out through a superposed residual channel attention block RCAB and a space attention module SA; in the RCAB, channel features are extracted through a global context module (GCB); performing reconstruction based on the depth features through the reconstruction convolution layer to obtain super-resolution image frames of the low-resolution image frames; and synthesizing all the super-resolution image frames into a super-resolution video.

Description

technical field [0001] The present invention relates to the technical field of video processing, and in particular, to a video processing method, device and storage medium based on a super-resolution network. Background technique [0002] High-definition video, such as 8k video shot by 8k camera, is not suitable for real-time transmission due to its large size. Therefore, before transmitting 8k video, the 8k video will be compressed. The specific processing method can be interlaced compression (for example, only odd lines or even lines). pixels), so the 8k video will be compressed into 4k video, and the video viewing terminal needs to restore the 4k video. In addition, scenarios of converting low-resolution videos to high-resolution videos are often encountered in real life, such as restoring old photos and videos. [0003] The existing super-resolution network needs to train a deep network structure in order to ensure the super-clear quality of the image, and the data proc...

AI Technical Summary

Problems solved by technology

[0005] The main purpose of the present invention is to provide a video processing method, device and computer-readable storage medium based on a super-resolution network, aiming to solve the problem that the super-resolution network in the prior art needs to train a deep network structure to ensure the ultra-clear image quality , the data processing speed is very slow, and it is not suitable for the problem of video super-resolution processing

Images