Performance of entropy-constrained reflected residual vector quantizer for generalized Gaussian sources

W. A.H. Mousa; M. A.U. Khan

doi:10.1109/DSPWS.2002.1231140

Performance of entropy-constrained reflected residual vector quantizer for generalized Gaussian sources

W. A.H. Mousa
, M. A.U. Khan

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

Abstract

Reflected residual vector quantization (RRVQ) was introduced as an alternative design algorithm for Residual Vector Quantization (RVQ) a.k.a. Multistage Vector Quantization. The advantage of RRVQ lies in its optimal single-path search of large RVQ trees. Furthermore, since RRVQ provides a structured codebook its output entropy is lower than that of RVQ codebook. The entropy-constrained RRVQ (EC-RRVQ) codebooks were designed for memoryless Gaussian and Laplacian sources with improved rate-distortion performance as compared to the entropyconstrained RVQ (EC-RVQ). In this paper the performance of EC-RRVQ is studied for a wide class of generalized Gaussian density sources. Generalized Gaussian density function (GGD) is a density function similar to the normal density function, however, it is defined in terms of the mean, standard deviation, and shape parameter which controls the exponential rate of decay. It has been successfully used in modeling transformed image pixels and subhand coefficients. Simulation results demonstrate that improved performance can be expected for GGD sources.

Original language	English
Title of host publication	Proceedings of 2002 IEEE 10th Digital Signal Processing Workshop, DSP 2002 and 2nd Signal Processing Education Workshop, SPE 2002
Publisher	Institute of Electrical and Electronics Engineers Inc.
Pages	397-399
Number of pages	3
ISBN (Electronic)	0780381165, 9780780381162
DOIs	https://doi.org/10.1109/DSPWS.2002.1231140
State	Published - 2002

Publication series

Name	Proceedings of 2002 IEEE 10th Digital Signal Processing Workshop, DSP 2002 and 2nd Signal Processing Education Workshop, SPE 2002

Bibliographical note

Publisher Copyright:
© 2002 IEEE.

ASJC Scopus subject areas

Signal Processing

Access to Document

10.1109/DSPWS.2002.1231140

Cite this

Mousa, W. A. H., & Khan, M. A. U. (2002). Performance of entropy-constrained reflected residual vector quantizer for generalized Gaussian sources. In Proceedings of 2002 IEEE 10th Digital Signal Processing Workshop, DSP 2002 and 2nd Signal Processing Education Workshop, SPE 2002 (pp. 397-399). Article 1231140 (Proceedings of 2002 IEEE 10th Digital Signal Processing Workshop, DSP 2002 and 2nd Signal Processing Education Workshop, SPE 2002). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/DSPWS.2002.1231140

Mousa, W. A.H. ; Khan, M. A.U. / Performance of entropy-constrained reflected residual vector quantizer for generalized Gaussian sources. Proceedings of 2002 IEEE 10th Digital Signal Processing Workshop, DSP 2002 and 2nd Signal Processing Education Workshop, SPE 2002. Institute of Electrical and Electronics Engineers Inc., 2002. pp. 397-399 (Proceedings of 2002 IEEE 10th Digital Signal Processing Workshop, DSP 2002 and 2nd Signal Processing Education Workshop, SPE 2002).

@inproceedings{d0bb024bd11b44fea049aee0c75ec955,

title = "Performance of entropy-constrained reflected residual vector quantizer for generalized Gaussian sources",

abstract = "Reflected residual vector quantization (RRVQ) was introduced as an alternative design algorithm for Residual Vector Quantization (RVQ) a.k.a. Multistage Vector Quantization. The advantage of RRVQ lies in its optimal single-path search of large RVQ trees. Furthermore, since RRVQ provides a structured codebook its output entropy is lower than that of RVQ codebook. The entropy-constrained RRVQ (EC-RRVQ) codebooks were designed for memoryless Gaussian and Laplacian sources with improved rate-distortion performance as compared to the entropyconstrained RVQ (EC-RVQ). In this paper the performance of EC-RRVQ is studied for a wide class of generalized Gaussian density sources. Generalized Gaussian density function (GGD) is a density function similar to the normal density function, however, it is defined in terms of the mean, standard deviation, and shape parameter which controls the exponential rate of decay. It has been successfully used in modeling transformed image pixels and subhand coefficients. Simulation results demonstrate that improved performance can be expected for GGD sources.",

author = "Mousa, \{W. A.H.\} and Khan, \{M. A.U.\}",

note = "Publisher Copyright: {\textcopyright} 2002 IEEE.",

year = "2002",

doi = "10.1109/DSPWS.2002.1231140",

language = "English",

series = "Proceedings of 2002 IEEE 10th Digital Signal Processing Workshop, DSP 2002 and 2nd Signal Processing Education Workshop, SPE 2002",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

pages = "397--399",

booktitle = "Proceedings of 2002 IEEE 10th Digital Signal Processing Workshop, DSP 2002 and 2nd Signal Processing Education Workshop, SPE 2002",

address = "United States",

}

Mousa, WAH & Khan, MAU 2002, Performance of entropy-constrained reflected residual vector quantizer for generalized Gaussian sources. in Proceedings of 2002 IEEE 10th Digital Signal Processing Workshop, DSP 2002 and 2nd Signal Processing Education Workshop, SPE 2002., 1231140, Proceedings of 2002 IEEE 10th Digital Signal Processing Workshop, DSP 2002 and 2nd Signal Processing Education Workshop, SPE 2002, Institute of Electrical and Electronics Engineers Inc., pp. 397-399. https://doi.org/10.1109/DSPWS.2002.1231140

Performance of entropy-constrained reflected residual vector quantizer for generalized Gaussian sources. / Mousa, W. A.H.; Khan, M. A.U.
Proceedings of 2002 IEEE 10th Digital Signal Processing Workshop, DSP 2002 and 2nd Signal Processing Education Workshop, SPE 2002. Institute of Electrical and Electronics Engineers Inc., 2002. p. 397-399 1231140 (Proceedings of 2002 IEEE 10th Digital Signal Processing Workshop, DSP 2002 and 2nd Signal Processing Education Workshop, SPE 2002).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Performance of entropy-constrained reflected residual vector quantizer for generalized Gaussian sources

AU - Mousa, W. A.H.

AU - Khan, M. A.U.

PY - 2002

Y1 - 2002

N2 - Reflected residual vector quantization (RRVQ) was introduced as an alternative design algorithm for Residual Vector Quantization (RVQ) a.k.a. Multistage Vector Quantization. The advantage of RRVQ lies in its optimal single-path search of large RVQ trees. Furthermore, since RRVQ provides a structured codebook its output entropy is lower than that of RVQ codebook. The entropy-constrained RRVQ (EC-RRVQ) codebooks were designed for memoryless Gaussian and Laplacian sources with improved rate-distortion performance as compared to the entropyconstrained RVQ (EC-RVQ). In this paper the performance of EC-RRVQ is studied for a wide class of generalized Gaussian density sources. Generalized Gaussian density function (GGD) is a density function similar to the normal density function, however, it is defined in terms of the mean, standard deviation, and shape parameter which controls the exponential rate of decay. It has been successfully used in modeling transformed image pixels and subhand coefficients. Simulation results demonstrate that improved performance can be expected for GGD sources.

AB - Reflected residual vector quantization (RRVQ) was introduced as an alternative design algorithm for Residual Vector Quantization (RVQ) a.k.a. Multistage Vector Quantization. The advantage of RRVQ lies in its optimal single-path search of large RVQ trees. Furthermore, since RRVQ provides a structured codebook its output entropy is lower than that of RVQ codebook. The entropy-constrained RRVQ (EC-RRVQ) codebooks were designed for memoryless Gaussian and Laplacian sources with improved rate-distortion performance as compared to the entropyconstrained RVQ (EC-RVQ). In this paper the performance of EC-RRVQ is studied for a wide class of generalized Gaussian density sources. Generalized Gaussian density function (GGD) is a density function similar to the normal density function, however, it is defined in terms of the mean, standard deviation, and shape parameter which controls the exponential rate of decay. It has been successfully used in modeling transformed image pixels and subhand coefficients. Simulation results demonstrate that improved performance can be expected for GGD sources.

UR - https://www.scopus.com/pages/publications/84954290831

U2 - 10.1109/DSPWS.2002.1231140

DO - 10.1109/DSPWS.2002.1231140

M3 - Conference contribution

AN - SCOPUS:84954290831

T3 - Proceedings of 2002 IEEE 10th Digital Signal Processing Workshop, DSP 2002 and 2nd Signal Processing Education Workshop, SPE 2002

SP - 397

EP - 399

BT - Proceedings of 2002 IEEE 10th Digital Signal Processing Workshop, DSP 2002 and 2nd Signal Processing Education Workshop, SPE 2002

PB - Institute of Electrical and Electronics Engineers Inc.

ER -

Mousa WAH, Khan MAU. Performance of entropy-constrained reflected residual vector quantizer for generalized Gaussian sources. In Proceedings of 2002 IEEE 10th Digital Signal Processing Workshop, DSP 2002 and 2nd Signal Processing Education Workshop, SPE 2002. Institute of Electrical and Electronics Engineers Inc. 2002. p. 397-399. 1231140. (Proceedings of 2002 IEEE 10th Digital Signal Processing Workshop, DSP 2002 and 2nd Signal Processing Education Workshop, SPE 2002). doi: 10.1109/DSPWS.2002.1231140

Performance of entropy-constrained reflected residual vector quantizer for generalized Gaussian sources

Abstract

Publication series

Bibliographical note

ASJC Scopus subject areas

Access to Document

Fingerprint

Cite this