Image coding has been the important problem in the computer world nowadays. Vector quantization (VQ) is a lossy image compression technique presented recently. It has the advantages of low bit rate and simple decoding method. However

the encoding phase of the full search method need much computation. In order to reduce the encoding time

a lot of fast search algorithms are presented in the literatures. However

lots of these methods cannot further reduce the bit rate. In this paper

a novel image encoding algorithm based on the fast correlation vector quantization (CVQ) is presented. The diagonal encoding sequence is adopted in this paper. During the encoding process

the correlation between the current processing block and the adjacent encoded blocks is used to predict the index of the input vector

thus both the average codeword searching range of each input vector and the bit rate are greatly reduced. Test results show that

compared with the image encoding algorithms based on the conventional full-search method

the partial distortion search method and the double test method

the proposed algorithm needs much shorter encoding time and lower bit rate

although the encoding quality of the proposed algorithm is a little degraded.