Cross-frame information fusion screen content video quality enhancement network

Abstract

The invention discloses a cross-frame information fusion screen content video quality enhancement network, which comprises a feature extraction module, a cross-frame information cross fusion module, a difference module, a reconstruction module and two 3 * 3 convolution modules, the feature extraction module performs feature extraction on the input frame; the cross-frame cross information fusion module is used for fusing the feature information; inputting the target frame into a 3 * 3 convolution module to obtain a feature map, and inputting feature information into another 3 * 3 convolution module to obtain a feature map FU; performing element subtraction operation on the FU sum to obtain high-frequency characteristics, and inputting the high-frequency characteristics into a difference module; the difference module recovers the high-frequency characteristic information; the reconstruction module is used for reconstructing the input features; and adding the reconstructed frame and the target frame to obtain an enhanced target frame. According to the method, the information of adjacent frames is fully mined, and the inter-frame information is fully utilized by adopting the cross-frame information cross fusion module, so that the subjective and objective quality of the compressed video is obviously enhanced.

Description

technical field [0001] The invention belongs to the technical field of image processing, in particular to a screen content video quality enhancement network for cross-frame information fusion. Background technique [0002] In recent years, with the rise of various online learning platforms and the live game industry, screen content video, a special type of video, has gradually attracted people's attention. Since video must be compressed using compression software during transmission, and the compressed video will produce compression artifacts, various quality enhancement schemes for compressed video have been extensively studied by industry and academia. [0003] However, there is currently no specific solution for quality enhancement of compressed screen content video. Therefore, this program proposes a compressed screen content video quality enhancement network based on cross-frame information fusion, abandoning the traditional optical flow-based and deformable volume-base...

AI Technical Summary

Problems solved by technology

However, accurate optical flow estimation is inherently challenging and time-consuming, so the quality enhancement effect of this invention is still limited.

[0007]With the increasing proportion of screen content video in the network, it is inevitable to compress and encode the screen content video to reduce the code during network transmission rate, but the reduction of the bit rate inevitably leads to the decline of the subjective and objective quality of the video

Images