Performance analysis of a vector quantization algorithm of image data

Summary form only given. This paper describes VQ techniques that use simple and fixed codebooks. The computation of the codebook is based on the statistical distribution of the intensity mean and standard deviation of vectors in the image. A formula is given for the expected signal to noise ratio versus image compression ratio. Non-adaptive codebook vectors have uniform entries based on the average intensity of vector entries mu . It is obvious that in the 8-bit imaging system, the range of mu is (0, 255). The adaptive approach uses the classification of vectors into two classes; non-edge vectors and edge vectors. A threshold based on the standard deviation sigma is used to establish this classification. Graphs show the results for both approaches.<
>