Vector quantization based on a binary search-like algorithm

Long Jhe Yan*, Shaw-Hwa Hwang, Shun Chieh Chang, Chi Jung Huang

*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

1 Scopus citations

Abstract

This paper presents an efficient binary search-like algorithm for vector quantization (VQ). The proposed algorithm adopts a tree-structured VQ with overlapped codewords (TSOC) to reduce computational complexity and enhance quantization quality. This algorithm uses overlapped codewords to expand the scope of the search path to traverse more appropriate codewords. To further evaluate computations at each stage of the proposed algorithm, both speech and images are considered. With codebook sizes of 256, 512 and 1024, the corresponding optimal computational savings for images are 85.16%, 90.04% and 93.46% respectively, compared with the FSVQ. For speech, the optimal computational savings reached 51.56% for a codebook size of 128. The results indicate that the proposed algorithm can save a significant number of computations, depending on the size of codebook.

Original languageEnglish
Title of host publicationFinal Program and Abstract Book - 4th International Symposium on Communications, Control, and Signal Processing, ISCCSP 2010
DOIs
StatePublished - 28 Jun 2010
Event4th International Symposium on Communications, Control, and Signal Processing, ISCCSP-2010 - Limassol, Cyprus
Duration: 3 Mar 20105 Mar 2010

Publication series

NameFinal Program and Abstract Book - 4th International Symposium on Communications, Control, and Signal Processing, ISCCSP 2010

Conference

Conference4th International Symposium on Communications, Control, and Signal Processing, ISCCSP-2010
CountryCyprus
CityLimassol
Period3/03/105/03/10

Keywords

  • Tree-structured VQ (TSVQ)
  • Tree-structured VQ with overlapped codewords (TSOC)
  • Triangle inequality elimination (TIE)
  • Vector quantization (VQ)

Fingerprint Dive into the research topics of 'Vector quantization based on a binary search-like algorithm'. Together they form a unique fingerprint.

Cite this