Skip to Main content Skip to Navigation
Habilitation à diriger des recherches

Quantification vectorielle algébrique : un outil performant pour la compression et le tatouage d'images fixes

Abstract : This manuscript describes twelve years of research activities in the Resarch Center for Automatic Control of Nancy in the field of image compression (real life and medical images), as well as in the field of watermarking associated to compression.
We emphasized the stage of quantification of the compression chain for which we proposed a method called " dead zone lattice vector quantization " (DZLVQ) associated with a multiresolution wavelet scheme, allowing to improve significantly the performances in terms of bit rate distortion trade-off and overall visual quality, with regard to the new standard JPEG2000 as well as to the reference algorithm SPIHT. We worked on three essential points and proposed every time the solutions to make the use of QVAZM realistic in a compression chain: the codebook vector labelling, the tuning of quantization parameters (scaling factor and deadzone size) and the bit allocation.
The major contribution of our works in the field of 3D medical imaging consisted in trying to open a way to lossy compression, still unthinkable some years ago for evident reasons of diagnosis. We extended successfully our DZLVQ algorithm to the case of volumetric medical images. Furthermore, we studied the impact of lossy compression computer-aided detection of solid lung nodules for which we were able to show a robustness to lossy compression even for high compression ratios (until 96:1).
Finally, the main contribution of our works in the .eld of watermarking concerns the development of combined compression/watermarking approaches which ended in the proposition of two methods based on DZLVQ principles. They are particularly attractive for the applications where the compression constitutes the main attack (or the main processing).
Document type :
Habilitation à diriger des recherches
Complete list of metadatas

Cited literature [124 references]  Display  Hide  Download

https://tel.archives-ouvertes.fr/tel-00201457
Contributor : Jean-Marie Moureaux <>
Submitted on : Saturday, December 29, 2007 - 3:34:34 PM
Last modification on : Friday, October 23, 2020 - 8:38:02 AM
Long-term archiving on: : Tuesday, April 13, 2010 - 3:50:09 PM

Identifiers

  • HAL Id : tel-00201457, version 1

Collections

Citation

Jean-Marie Moureaux. Quantification vectorielle algébrique : un outil performant pour la compression et le tatouage d'images fixes. Traitement du signal et de l'image [eess.SP]. Université Henri Poincaré - Nancy I, 2007. ⟨tel-00201457⟩

Share

Metrics

Record views

565

Files downloads

2618