Efficient Learned Lossless JPEG Recompression

TL;DR

This paper introduces a deep learning-based method for lossless JPEG image compression that achieves higher compression ratios and lower latency through a novel parallel conditional modeling architecture, with potential for high-quality lossy compression.

Contribution

The paper presents a Multi-Level Parallel Conditional Modeling architecture enabling parallel decoding and improved compression efficiency for JPEG images.

Findings

Achieves better compression ratios than previous state-of-the-art methods.

Attains a throughput of 57 FPS for 1080P images on NVIDIA T4 GPU.

Effective for high bit rate lossy JPEG compression with advantages over existing methods.

Abstract

JPEG is one of the most popular image compression methods. It is beneficial to compress those existing JPEG files without introducing additional distortion. In this paper, we propose a deep learning based method to further compress JPEG images losslessly. Specifically, we propose a Multi-Level Parallel Conditional Modeling (ML-PCM) architecture, which enables parallel decoding in different granularities. First, luma and chroma are processed independently to allow parallel coding. Second, we propose pipeline parallel context model (PPCM) and compressed checkerboard context model (CCCM) for the effective conditional modeling and efficient decoding within luma and chroma components. Our method has much lower latency while achieves better compression ratio compared with previous SOTA. After proper software optimization, we can obtain a good throughput of 57 FPS for 1080P images on NVIDIA T4 GPU. Furthermore, combined with quantization, our approach can also act as a lossy JPEG codec which has obvious advantage over SOTA lossy compression methods in high bit rate (bpp$>0.9$).