Joint spatial frequency quantization of zerotree wavelet video coding

Abstract

This paper proposes joint spatial-frequency quantization for a zerotree wavelet video encoder. Following temporal redundancy reduction via block motion estimation, the displaced frame difference signal is discrete wavelet transformed, zerotree encoded and optimally quantized. Improved performance is obtained compared to H.263 and systems nor employing joint quantization, while at the same time preserving spatial and frequency scalability.