Effects of Codebook Sizes, Codeword Dimensions, and Colour Spaces on Retrieval Performance of Image Retrieval Using Vector Quantization

Abstract

Recently, we have proposed an image indexing and retrieval technique that is based on vector quantization. We have already shown that this technique is more effective than the traditional colour-based techniques. Some factors that must be decided during the implementation of the proposed techniques are codebook size, codeword dimension and colour space. In this paper, we investigate how these factors may affect the retrieval performances of the proposed technique.