Adaptive wavelet packet basis for entropy-constrained lattice vector quantizer (ECLVQ)

摘要

In this paper, the search algorithm for the best wavelet basis/quantization pair is proposed, while specifying the source coding strategy as entropy-constrained lattice vector quantization (ECLVQ). The presented approach differs from previous works since the wavelet basis found is quasi-optimal but the quantizer set is optimally chosen. In addition, the source coding is extended to quantization plus entropy coding, and the computational complexity is reduced by using the estimation formulas instead of measuring the quantization distortion.