This paper proposes a novel variable-rate optimal daptive scalar quantization algorithm, and reports preliminay results on its usage in quantization for image ocmpression. The heart of the methdology is a Dynamic programming algorithm for fast variable-rate codebook-constrained optimal scalar quantizer design. The present algorithm is the variable-rate counterpart of a recent Dp algorithm for fixed-rate codebook-constrained optimal scalar uantizer design. te two algorithms have a common trait: both optimiz the design subject to the constraint thatthe quantizanon levels be chosen from a hypercodebook of candidate levels. This restiiction is meaningless in the cntext of fixed quantizder design, but very useul in the context of adaptive quantization, where a description of the re-designed codebook must also be communicated to the decoder. The present paper describes the variable-rate algorithm, reportts some rate-distorition performance results, and compares the performance fote variable-rae algorithm versus the fixed-rate algorithm.
展开▼
