Lossless image compression using reversible integer wavelet transforms and convolutional neural networks

Research output: Chapter in Book/Report/Conference proceedingConference contribution

8 Scopus citations

Abstract

In this work we introduce a lossless compression framework which incorporates convolutional neural networks (CNN) for wavelet subband prediction. A CNN is trained to predict detail coefficients from corresponding approximation coefficients, prediction error is then coded in place of wavelet coefficients. At decompression an identical CNN is used to reproduce the prediction and combine with the decoded residuals for perfect reconstruction of wavelet subbands.

Original languageEnglish (US)
Title of host publicationProceedings - DCC 2018
Subtitle of host publication2018 Data Compression Conference
EditorsAli Bilgin, James A. Storer, Joan Serra-Sagrista, Michael W. Marcellin
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages395
Number of pages1
ISBN (Electronic)9781538648834
DOIs
StatePublished - Jul 19 2018
Event2018 Data Compression Conference, DCC 2018 - Snowbird, United States
Duration: Mar 27 2018Mar 30 2018

Publication series

NameData Compression Conference Proceedings
Volume2018-March
ISSN (Print)1068-0314

Other

Other2018 Data Compression Conference, DCC 2018
CountryUnited States
CitySnowbird
Period3/27/183/30/18

Keywords

  • cnn
  • compression
  • convolutional neural network
  • deep learning
  • wavelets

ASJC Scopus subject areas

  • Computer Networks and Communications

Fingerprint Dive into the research topics of 'Lossless image compression using reversible integer wavelet transforms and convolutional neural networks'. Together they form a unique fingerprint.

Cite this