You Can Mask More For Extremely Low-Bitrate Image Compression
Anqi Li, Feng Li, Jiaxin Han, Huihui Bai, Runmin Cong, Chunjie Zhang,, Meng Wang, Weisi Lin, Yao Zhao

TL;DR
This paper introduces a novel end-to-end framework combining masked image modeling and learned image compression to significantly improve extremely low-bitrate image compression, focusing on structure and texture components.
Contribution
It proposes a dual-adaptive masking strategy and a unified masked compression model that enhance low-bitrate image compression by leveraging pre-trained MAE and LIC techniques.
Findings
Outperforms state-of-the-art methods in rate-distortion performance
Achieves better visual quality at very low bitrates
Enhances downstream application performance
Abstract
Learned image compression (LIC) methods have experienced significant progress during recent years. However, these methods are primarily dedicated to optimizing the rate-distortion (R-D) performance at medium and high bitrates (> 0.1 bits per pixel (bpp)), while research on extremely low bitrates is limited. Besides, existing methods fail to explicitly explore the image structure and texture components crucial for image compression, treating them equally alongside uninformative components in networks. This can cause severe perceptual quality degradation, especially under low-bitrate scenarios. In this work, inspired by the success of pre-trained masked autoencoders (MAE) in many downstream tasks, we propose to rethink its mask sampling strategy from structure and texture perspectives for high redundancy reduction and discriminative feature representation, further unleashing the potential…
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Code & Models
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsAdvanced Image Processing Techniques · Image and Signal Denoising Methods · Image Enhancement Techniques
Methodsfail · Masked autoencoder · Mutual Information Machine/Mask Image Modeling
