Region-of-interest-based video coding for video conference applications

Abstract : This PhD. thesis addresses the problem of region-of-interest-based video coding and deals with improving the coding efficiency in High Efficiency Video Coding. We propose both accurate rate distortion modeling approaches, and also region-of-interest-based rate control methods adapted for High Efficiency Video Coding. In the first part, we propose new rate-distortion models for High Efficiency Video Coding at coding unit level. Proposed modeling takes into account content characteristics and encoder features. In a first proposition models are based on spatial dependencies between pixels of a coding unit while in a second proposition statistical characteristics of the data are used to derive more efficient models. We show the benefits that can be drawn from using content based rate-distortion modeling. A good fitting of transform coefficients per unit gives us important gains in coding efficiency. In the second part, we propose novel rate control algorithms for High Efficiency Video Coding that introduces region-of-interest concept. Performing bit allocation per region and computing quantization parameter independently per units of various importance levels, help improving budget partitioning over regions of different interest. This can be useful in many applications where region-based processing of the frame is required such as videoconferencing systems. The proposed methods show an improvement in the quality of the region-of-interest while the budget constraint is respected.
Complete list of metadatas

Cited literature [92 references]  Display  Hide  Download

https://hal-imt.archives-ouvertes.fr/tel-01410517
Contributor : Cagnazzo Marco <>
Submitted on : Tuesday, December 6, 2016 - 4:40:59 PM
Last modification on : Friday, December 9, 2016 - 1:00:56 AM
Long-term archiving on : Tuesday, March 21, 2017 - 3:37:37 PM

Identifiers

  • HAL Id : tel-01410517, version 1

Collections

Citation

Marwa Meddeb. Region-of-interest-based video coding for video conference applications. Signal and Image processing. Telecom ParisTech, 2016. English. ⟨tel-01410517⟩

Share

Metrics

Record views

99

Files downloads

705