Constrained quantization algorithm for color images

Author

Yan, Jingqi ; Yang, Xin ; Shi, Pengfei

Author_Institution

Inst. of Image Process. & Pattern Recognition, Shanghai Jiao Tong Univ., China

Volume

1

fYear

2001

fDate

6/23/1905 12:00:00 AM

Firstpage

910

Abstract

In this paper, we discuss two kinds of VQ algorithms. One is based on minimizing the total variance and the other is based on minimizing the maximum deviation. The algorithms of the first kind better reflect the overall fit, but may discount large, but highly localized, deviations. Those of the second kind provide absolute distance bounds that are a useful error guarantee, but may be overly sensitive to any noise that might be present in the original models. A new algorithm, combining the two criteria, is presented in this paper. It not only improves the total variance, but also provides a useful maximum error guarantee. The experiments indicate the new quantizer is a better choice in some practical operations

Keywords

image coding; image colour analysis; minimisation; vector quantisation; VQ algorithms; absolute distance bounds; color images; constrained quantization algorithm; maximum deviation; maximum error guarantee; noise; total variance; Acceleration; Color; Costs; Graphics; Hardware; Identity-based encryption; Image processing; Image storage; Pattern recognition; Vector quantization;

fLanguage

English

Publisher

ieee

Conference_Titel

Image Processing, 2001. Proceedings. 2001 International Conference on

Print_ISBN

0-7803-6725-1

Type

conf

DOI

10.1109/ICIP.2001.959194

Filename

959194