CLASSIFIED VECTOR QUANTIZATION USING VARIANCE CLASSIFIER AND MAXIMUM-LIKELIHOOD CLUSTERING

A classified vector image quantizer is proposed here. The algorithm em ploys a one-feature variance classifier. This classifier has good prop erties as it sorts the classes by its entropy contents. Then every cla ss is clustered using the mixture maximum likelihood criterion instead of the Euclidean distance. This shows that the number of clusters req uired to represent any class can be determined. It also provides bette r clustering by emphasizing on fitting a model of a mixture of Gaussia n distributions by the data. Impressive results regarding the image qu ality and bit rates are obtained when the algorithm is applied to imag e compression.