STOCHASTIC VECTOR QUANTIZATION OF IMAGES

One of the most important steps in the vector quantization of images i s the design of the codebook. The codebook is generally designed using the LBG algorithm, that is in essence a clustering algorithm which us es a large training set of empirical data that is statistically repres entative of the image to be quantized. The LBG algorithm, although qui te effective for practical applications, is computationally very expen sive and the resulting codebook has to be recalculated each time the t ype of image to be encoded changes. One alternative to the generation of the codebook, called stochastic vector quantization, is presented i n this paper. Stochastic vector quantization (SVQ) is based on the gen eration of the codebook according to some previous model defined for t he image to be encoded. The well-known AR model has been used to model the image in the current implementations of the technique, and has sh own good performance in the overall scheme. To show the merit of the t echnique in different contexts, stochastic vector quantization is disc ussed and applied to both pixel-based and segmentation-based image cod ing schemes. (C) 1997 Elsevier Science B.V.