Minimum distortion clustering technique for orthogonal polynomials transform vector quantizer

Authors:
R. Krishnamoorthy;J. Kalpana
Affiliations:
Anna University of Technology, Trichirappalli, Tamil Nadu, India;Bharathidasan Inst. of Tech., Trichirappalli, Tamil Nadu, India
Venue:
Proceedings of the 2011 International Conference on Communication, Computing & Security
Year:
2011

Citing 4
Cited 0

Self-Organizing Maps

Self-Organizing Maps
Transform coding of monochrome images with a statistical design of experiments approach to separate noise

Pattern Recognition Letters
A new integer image coding technique based on orthogonal polynomials

Image and Vision Computing
Face detection using quantized skin color regions merging andwavelet packet analysis

IEEE Transactions on Multimedia

Quantified Score

Hi-index	0.00

Visualization

Abstract

Even though Lloyd's algorithm is all pervasive in the design of clustering algorithms like vector quantizers (VQ), it's drawbacks are that the codebook generation time is size dependent and that the initial codebook is not a minimum distortion one. The proposed clustering technique has two objectives-the first objective is to design a codebook in a time that is size-independent and the second one is to produce a minimum distortion, high-quality initial codebook. If the image is transformed into a domain in which its elements are uncorrelated, VQ performs better and so Orthogonal Polynomials Transform (OPT) is applied earlier to vector quantization. The dimensionality reduction coupled with pruning of transform coefficients is indeed an attractive feature of this transform vector quantizer (TVQ). The performance of the proposed technique is observed in the vector space of OPT coefficients and has been compared with other VQs. Experiments done reveal the superiority of the proposed technique in terms of increased PSNR. Results also show that the Mean Squared Error (MSE) decays drastically within the first two iterations with the proposed technique.