IMAGE COMPRESSION VIA VECTOR QUANTIZATION WITH VARIABLE DIMENSION.

Yasuo Matsuyama

Research output: Chapter in Book/Report/Conference proceedingConference contribution

5 Citations (Scopus)

Abstract

Novel vector-quantization algorithms are given and applied to image compression. These algorithms extend the conventional algorithms which is a direct generalization of the classical Lloyd-Max type. The quantizer generates subregions of convex polygons with different sizes. Novel size of the convex polygon is chosen so that the total distortion of the training or source image to a given codebook is minimized. Experiments on 8-b/pixel images show that the optimization of the variable-region (or variable-dimension) vector quantization subdivides the original rectilinear pixel array into versatile convex polygons allowing slant edges.

Original languageEnglish
Title of host publicationUnknown Host Publication Title
Place of PublicationNew York, NY, USA
PublisherIEEE
Pages423-427
Number of pages5
Publication statusPublished - 1987
Externally publishedYes

Fingerprint

Vector quantization
Image compression
Pixels
Experiments

ASJC Scopus subject areas

  • Engineering(all)

Cite this

Matsuyama, Y. (1987). IMAGE COMPRESSION VIA VECTOR QUANTIZATION WITH VARIABLE DIMENSION. In Unknown Host Publication Title (pp. 423-427). New York, NY, USA: IEEE.

IMAGE COMPRESSION VIA VECTOR QUANTIZATION WITH VARIABLE DIMENSION. / Matsuyama, Yasuo.

Unknown Host Publication Title. New York, NY, USA : IEEE, 1987. p. 423-427.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Matsuyama, Y 1987, IMAGE COMPRESSION VIA VECTOR QUANTIZATION WITH VARIABLE DIMENSION. in Unknown Host Publication Title. IEEE, New York, NY, USA, pp. 423-427.
Matsuyama Y. IMAGE COMPRESSION VIA VECTOR QUANTIZATION WITH VARIABLE DIMENSION. In Unknown Host Publication Title. New York, NY, USA: IEEE. 1987. p. 423-427
Matsuyama, Yasuo. / IMAGE COMPRESSION VIA VECTOR QUANTIZATION WITH VARIABLE DIMENSION. Unknown Host Publication Title. New York, NY, USA : IEEE, 1987. pp. 423-427
@inproceedings{5edde71b05794e7bb172b182af5d5efa,
title = "IMAGE COMPRESSION VIA VECTOR QUANTIZATION WITH VARIABLE DIMENSION.",
abstract = "Novel vector-quantization algorithms are given and applied to image compression. These algorithms extend the conventional algorithms which is a direct generalization of the classical Lloyd-Max type. The quantizer generates subregions of convex polygons with different sizes. Novel size of the convex polygon is chosen so that the total distortion of the training or source image to a given codebook is minimized. Experiments on 8-b/pixel images show that the optimization of the variable-region (or variable-dimension) vector quantization subdivides the original rectilinear pixel array into versatile convex polygons allowing slant edges.",
author = "Yasuo Matsuyama",
year = "1987",
language = "English",
pages = "423--427",
booktitle = "Unknown Host Publication Title",
publisher = "IEEE",

}

TY - GEN

T1 - IMAGE COMPRESSION VIA VECTOR QUANTIZATION WITH VARIABLE DIMENSION.

AU - Matsuyama, Yasuo

PY - 1987

Y1 - 1987

N2 - Novel vector-quantization algorithms are given and applied to image compression. These algorithms extend the conventional algorithms which is a direct generalization of the classical Lloyd-Max type. The quantizer generates subregions of convex polygons with different sizes. Novel size of the convex polygon is chosen so that the total distortion of the training or source image to a given codebook is minimized. Experiments on 8-b/pixel images show that the optimization of the variable-region (or variable-dimension) vector quantization subdivides the original rectilinear pixel array into versatile convex polygons allowing slant edges.

AB - Novel vector-quantization algorithms are given and applied to image compression. These algorithms extend the conventional algorithms which is a direct generalization of the classical Lloyd-Max type. The quantizer generates subregions of convex polygons with different sizes. Novel size of the convex polygon is chosen so that the total distortion of the training or source image to a given codebook is minimized. Experiments on 8-b/pixel images show that the optimization of the variable-region (or variable-dimension) vector quantization subdivides the original rectilinear pixel array into versatile convex polygons allowing slant edges.

UR - http://www.scopus.com/inward/record.url?scp=0023534203&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0023534203&partnerID=8YFLogxK

M3 - Conference contribution

SP - 423

EP - 427

BT - Unknown Host Publication Title

PB - IEEE

CY - New York, NY, USA

ER -