A generalized multiple projection axes method for fast encoding of vector quantization

Authors:
Zhibin Pan;Koji Kotani;Tadahiro Ohmi
Affiliations:
New Industry Creation Hatchery Center, Tohoku University, Aza-aoba 05, Aramaki, Aoba-ku, Sendai 980-8579, Japan;Department of Electronic Engineering, Graduate School of Engineering, Tohoku University, Aza-aoba 05, Aramaki, Aoba-ku, Sendai 980-8579, Japan;New Industry Creation Hatchery Center, Tohoku University, Aza-aoba 05, Aramaki, Aoba-ku, Sendai 980-8579, Japan
Venue:
Pattern Recognition Letters
Year:
2005

Citing 5
Cited 1

An efficient approximation-elimination algorithm for fast nearest-neighbor search based on a spherical distance coordinate formulation

Pattern Recognition Letters
Equal-average hyperplane partitioning method for vector quantization of image data

Pattern Recognition Letters
A fast vector quantization encoding algorithm using multiple projection axes

Signal Processing
A fast full search equivalent encoding method for vector quantization by using appropriate features

ICME '03 Proceedings of the 2003 International Conference on Multimedia and Expo - Volume 1
Application of Principal Component Analysis to Multikey Searching

IEEE Transactions on Software Engineering

Fuzzy vector quantization with the particle swarm optimization: A study in fuzzy granulation-degranulation information processing

Signal Processing

Quantified Score

Hi-index	0.10

Visualization

Abstract

The encoding process of vector quantization (VQ) is computationally very expensive due to a lot of k-dimensional Euclidean distance computations. In order to speed up VQ's encoding process, it is very effective to use a computationally inexpensive distance estimation first to try to reject a candidate codeword instead of an immediate actual distance computation. If a successful rejection is achieved, the computational burden can be reduced because the actual distance computation becomes unnecessary. A very search-efficient VQ encoding method by using multiple projection axes has already been developed in the previous work, which is a general version of the central axis. In this paper, a further generalized version of this previous work is proposed, which can completely remove the constraints for selecting the projection axis in a diagonally symmetric way as required by the previous work. Meanwhile, a theoretical criterion of how to select an optimal projection axis for a candidate codeword is also given. Furthermore, in order to use the generalized multi projection axes simultaneously, the energy accumulation property in an orthogonal space is integrated. Experimental results confirmed the effectiveness of the proposed method.