A new wavelet-based image coding using embedded vector quantizer

Author

Abdelwahab, Ahmed A. ; Saad, E.M. ; Gaber, Sarnir

Author_Institution

Fac. of Eng., Helwan Univ., Cairo, Egypt

Volume

2

fYear

2002

fDate

2002

Firstpage

898

Abstract

In this paper, only three scales of wavelet decompositions having ten different bands of coefficients are used for image compression. The DC band coefficients are transmitted using the gain.-shape vector quantizer. The nine different ac bands are vector quantized by three new proposed so called embedded vector quantizers. The purpose of the embedded vector quantizer is to send one address for each of the three codebooks of the same spatial three AC bands resulting in saving a lot of search time (95.24% of full search time) and in reducing the bit rate. Several scenarios with different bit rates as low as 0.109 bpp are followed giving PSNRs as high as 28-29 dB.

Keywords

discrete wavelet transforms; image coding; transform coding; vector quantisation; AC bands; DC band coefficients; PSNR; bit rate reduction; codebooks; computer simulation results; discrete wavelet transform decomposition; embedded VQ; embedded vector quantizer; gain.-shape vector quantizer; image compression; linear phase two-channel fifter banks; search time; wavelet decomposition; wavelet-based image coding; Bit rate; Discrete wavelet transforms; Filter bank; Fourier transforms; Image coding; Nonlinear filters; Pixel; Signal analysis; Wavelet analysis; Wavelet packets;

fLanguage

English

Publisher

ieee

Conference_Titel

Electrical and Computer Engineering, 2002. IEEE CCECE 2002. Canadian Conference on

ISSN

0840-7789

Print_ISBN

0-7803-7514-9

Type

conf

DOI

10.1109/CCECE.2002.1013062

Filename

1013062