Evaluation of Clustering Algorithms for Vector Quantization in Practical Usage

Abstract

Image coding technologies are essential to use communication channels effectively. We have been studied vector quantization, because it does not cause deterioration in image quality in a high compression region and also has a small computational cost for image coding. The performance of vector quantization depends on a code book that is constructed using clustering algorithms. In this study, we evaluated popularly used typical clustering algorithms in terms of practical usage. Computational experiments showed that there is no notable difference in performance of the clustering algorithms in practical usage of vector quantization.