Wavelet-based image coder with entropy-constrained lattice vectorquantizer (ECLVQ)

摘要

This paper presents an algorithm that jointly optimizes a latticenvector quantizer (LVQ) and an entropy coder in a subband coding at allnranges of bit rate. Estimation formulas for both entropy and distortionnof lattice quantized subband images are derived. From these estimates,nwe then develop dynamic algorithm optimizing the LVQ and entropy coderntogether for a given entropy rate. Compared to previously reportednmin-max approaches, or approaches using asymptotic distortion bounds,nthe approach reported here quickly designs a highly accurate optimalnentropy-constrained LVQ. The corresponding wavelet-based image codernalso has better coding performance compared to other subband coders thatnuse entropy-constrained LVQ, especially at low bit rates