A new variable-rate finite-state vector quantization for image compression

In this paper, a new variable-rate finite-state vector quantization (VRFSVQ) for image compression is presented. The VRFSVQ codec system consists of a tree-structured vector quantizer via the M-algorithm for searching, called M-TSVQ, a next-state generator which is specified by using index-matching criterion, and a variable-length codec. From the simulation results, a very low bit-rate image codec with good image quality and low encoding complexity may be achieved.