Home // IMMM 2023, The Thirteenth International Conference on Advances in Information Mining and Management // View article

A Swin Transformer Based Restoration Scheme for VVC Compressed Images

Authors:
Zhenchao Ma
Yixiao Wang
Hamid Reza Tohidypour
Panos Nasiopoulos
Victor C. M. Leung

Keywords: image restoration; VVC; vision transformer; multi scale window; artifacts reduction.

Abstract:
The Versatile Video Coding (VVC) standard is shown to significantly outperform the High Efficiency Video Coding (HEVC), the previous compression standard image/video codecs. More complex structures and advanced prediction techniques are behind this improved performance, leading to reduced visual artifacts. Deep learning-based image restoration algorithms have been proposed and are increasingly used for further reducing VVC generated artifacts. In this paper, we propose a Swin Transformer based image restoration model for VVC compression artifacts reduction that employs a self-attention mechanism to explore both global and local features to better understand the relation between existing and missing information. Performance evaluations showed that our proposed method outperforms existing state-of-the-art approaches yielding 0.884 dB quality improvement or 15.95% bitrate savings.

Pages: 1 to 5

Copyright: Copyright (c) IARIA, 2023

Publication date: June 26, 2023

Published in: conference

ISSN: 2326-9332

ISBN: 978-1-68558-072-8

Location: Nice, France

Dates: from June 26, 2023 to June 30, 2023