Skip to Main content Skip to Navigation
Theses

New Trends in High Definition Video Compression - Application to Multiple Description Coding

Abstract : The framework of the thesis is a wavelet-based video coder. Fully scalable, this video encoder is based on a lifted motion-compensated wavelet transform. The first challenge was to reduce the cost of the motion vectors, which can be prohibitive at low bit-rates, by quantizing with losses the vectors. This method has been applied to the H.264 coder. The goal is to find the optimal bit-rates for the motion vectors and for the temporal wavelet coefficients in order to minimize the total distortion. A theoretical distortion model has thus been established, and an optimal bit-rate allocation has been realized. The influence of some badly estimated motion vectors on the motion-compensated wavelet transform has also been minimized. The steps of the lifting scheme have been closely adapted to the energy of the motion. To deal with the problems of efficient video transmission over noisy channels,Multiple Description Coding (MDC) has been explored. The framework is a balanced MDC scheme for scan-based wavelet transform video coding. A focus is done on the joint decoding of descriptions received at decoder and corrupted by noise. The challenge is to reconstruct a central signal with a distortion as small as possible using the knowledge of the probability density function of the descriptions, by two different algorithms. Distributed video coding has also been explored.
Complete list of metadatas

Cited literature [228 references]  Display  Hide  Download

https://tel.archives-ouvertes.fr/tel-00451684
Contributor : Estelle Nivault <>
Submitted on : Friday, January 29, 2010 - 4:43:17 PM
Last modification on : Monday, October 12, 2020 - 10:30:32 AM
Long-term archiving on: : Friday, June 18, 2010 - 5:33:26 PM

Identifiers

  • HAL Id : tel-00451684, version 1

Collections

Citation

Marie Andrée Agostini. New Trends in High Definition Video Compression - Application to Multiple Description Coding. Signal and Image processing. Université Nice Sophia Antipolis, 2009. English. ⟨tel-00451684⟩

Share

Metrics

Record views

320

Files downloads

1657