Learning-based lossless light field compression

Milan Stepanov, M. Umair Mukati, Giuseppe Valenzise, Søren Forchhammer, Frederic Dufaux

Research output: Chapter in Book/Report/Conference proceedingArticle in proceedingsResearchpeer-review

35 Downloads (Pure)


We propose a learning-based method for lossless light field compression. The approach consists of two steps: first, the view to be compressed is synthesized based on previously decoded views; then, the synthesized view is used as a context to predict probabilities of the residual signal for adaptive arithmetic coding. We leverage recent advances in deep-learning-based view synthesis and generative modeling. Specifically, we evaluate two strategies for entropy modeling: a fully parallel probability estimation, where all pixel probabilities are estimated simultaneously; and a partially auto-regressive estimation, in which groups of pixels are predicted sequentially. Our results show that the latter approach provides the best coding gains compared to the state of the art, while keeping the computational complexity competitive.
Original languageEnglish
Title of host publicationProceedings of 23rd IEEE International Workshop on Multimedia Signal Processing
Number of pages6
Publication date2021
ISBN (Print)9781665432887
Publication statusPublished - 2021
Event2021 IEEE 23rd International Workshop on Multimedia Signal Processing - Hybrid event, Tampere, Finland
Duration: 6 Oct 20218 Oct 2021
Conference number: 23


Conference2021 IEEE 23rd International Workshop on Multimedia Signal Processing
LocationHybrid event
Internet address


  • Light field
  • Lossless Coding
  • Deep learning


Dive into the research topics of 'Learning-based lossless light field compression'. Together they form a unique fingerprint.

Cite this