Gamma rate theory for causal rate control in source coding and video coding

Jun Xu, Dapeng Wu*

*Corresponding author for this work

Research output: Journal Publications and ReviewsRGC 21 - Publication in refereed journalpeer-review

1 Citation (Scopus)

Abstract

The rate distortion function in information theory provides performance bounds for lossy source coding. However, it is not clear how to causally encode a Gaussian sequence under rate constraints while achieving R-D optimality. This problem has significant implications in the design of rate control for video communication. To address this problem, we take distortion fluctuation into account and develop a new theory, called gamma rate theory, to quantify the trade-off between rate and distortion fluctuation. The gamma rate theory implies that, to evaluate the performance of causal rate controls in source coding, the traditional R-D metric needs to be replaced by a new GRD metric. The gamma rate theory identifies the trade-off between quality fluctuation and bandwidth, which is not known previously. To validate the gamma rate theory, we design a rate control algorithm for video coding; our experimental results demonstrate the utility of the gamma rate theory in video coding.
Original languageEnglish
Pages (from-to)1886-1893
JournalJournal of Visual Communication and Image Representation
Volume25
Issue number8
DOIs
Publication statusPublished - Nov 2014
Externally publishedYes

Bibliographical note

Publication details (e.g. title, author(s), publication statuses and dates) are captured on an “AS IS” and “AS AVAILABLE” basis at the time of record harvesting from the data source. Suggestions for further amendments or supplementary information can be sent to [email protected].

Research Keywords

  • Causal rate control
  • Gamma rate function
  • Network friendliness
  • Rate distortion function
  • Rate gamma function
  • Source coding
  • Video coding
  • Viewer friendliness

Fingerprint

Dive into the research topics of 'Gamma rate theory for causal rate control in source coding and video coding'. Together they form a unique fingerprint.

Cite this