Home About Login Current Archives Announcements Editorial Board
Submit Now For Authors Call for Submissions Statistics Contact
Home > Archives > Volume 20, No 11 (2022) > Article

DOI: 10.14704/nq.2022.20.11.NQ66270

VIDEO COMPRESSION USING RATE-DISTORTION OPTIMIZATION-BASED CHANNEL ATTENTION NETWORK

B.Shravan Kumar, Dr. V.Usha Shree, Dr. Sumagna Patnaik

Abstract

In this study, we provide a hybrid video compression methodology focused on perceptual quality optimization. Multi-scale optimal coding methods are used in the proposed framework, which is based on the newly released Versatile Video Coding (VVC). From the coding unit level to the video sequence level, specific attention has been paid to three key areas in order to significantly enhance compression efficiency. In order to remove block artefacts, we first suggest a block-level rate-distortion optimization (RDO) approach. Post-processing of each compressed image is then handled by convolutional neural networks with frame-level perceptual quality optimizations that use a channel attention method to capture and restore the key information in subjective assessment. To get the greatest balance between quality and bit rate, it is important to treat bit allocation as a dynamic programming problem. According to experimental data, the validation set in the video track of the proposed technique by CLIC-2021 achieved an MS-SSIM of 0.98658.

Keywords

Versatile Video Coding, rate-distortion optimization, and channel attention network are other related terms.

Full Text

PDF

References

?>