Improving Quantization Matrices for Image Coding by Machine Learning

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

2 Citations (Scopus)

Abstract

We investigate the generation of quantization matrices for image coding in the scenario to balance compression ratio and quality. We make use of machine learning to train and determine those quantization matrices that can achieve the best compression ratio while reaching the quality settings. By introducing the trainable parameters and considering the impact of the quantization module on task performance and compression ratio, the DCT and quantization modules are jointly optimized to minimize the total coding cost. We evaluate the well-trained quantization matrices under various quality settings of JPEG. The results indicate that the proposed scheme can be combined with quality settings to consistently achieve better compression performance.

Original languageEnglish
Title of host publicationICDSP 2022 - 2022 6th International Conference on Digital Signal Processing
PublisherAssociation for Computing Machinery
Pages115-119
Number of pages5
ISBN (Electronic)9781450395809
DOIs
Publication statusPublished - 25 Feb 2022
Event6th International Conference on Digital Signal Processing, ICDSP 2022 - Virtual, Online, China
Duration: 25 Feb 202227 Feb 2022

Publication series

NameACM International Conference Proceeding Series
VolumePar F180471

Conference

Conference6th International Conference on Digital Signal Processing, ICDSP 2022
Country/TerritoryChina
CityVirtual, Online
Period25/02/2227/02/22

Keywords

  • DCT
  • JPEG
  • PyTorch
  • Quantization Matrix
  • SSIM

Fingerprint

Dive into the research topics of 'Improving Quantization Matrices for Image Coding by Machine Learning'. Together they form a unique fingerprint.

Cite this