Adaptive vector quantization based on quality criterion using Hosaka plot

In this paper, adaptive vector quantization based on the quality of the picture is suggested. In this method the codebook size is kept variable. New code vectors are added such that the quality of the image is improved compared to the fixed codebook. Code vectors are not at added at random from a supercodebook, which is normally done in adaptive vector quantization but only selective code vectors are added such that each code vector is useful for the improvement in the quality. This keeps the codebook size limited. Hence the time required for the compression or the memory requirement is not increased due to adaptive codebook yet the quality is improved. Therefore real time implementation is possible using special hardware. This algorithm is successful for any type of image. With the addition of 25% code vectors the maximum quality was improved