Video Compression using Adaptive Wavelet Packets and DPCM

Abstract

This paper presents a video compression algorithm based on the adaptive wavelet packet decomposition and DPCM methods, in which, motion compensation is not used explicitly, but the effects of motion are accounted for through the low-high temporal subbanding and the DPCM procedure. The coefficients of high temporal subbands, initiallowpass sub bands and the difference of low subbands from the DPCM loop are decomposed by adaptive wavelet packets based on a rate-distortion criterion, such that the subbands of the wavelet packets are efficiently quantized by a hybrid Scalar/Pyramidal Lattice Vector quantizer. We implement two versions of the proposed coding scheme based on a rate constraint and a quality (distortion) constraint. When the rate is constrained, all frames in a "group" are processed as a unit; our objective then is to minimize the overall distortion under the rate constraint specified for that group of frames. When quality constraint is imposed, pairs of frames are sequentially encoded; our goal in this case is to ensure that the distortion in each pair of frames remains below a prespecified value and to achieve this with a minimum bit rate used to represent that pair of frames. The results of the proposed algorithm compare favorably with those of 3D subband coding and traditional motion compensated video coding techniques. Although the encoding computational complexity of this scheme is higher than that of the existing 3D subband method, it is suitable for parallel implementation.

Efficient wavelet-based video coding

“…SPIHT [31]) significantly for textured images in terms of rate-distortion performance. A similar development in the area of video compression may be observed for 3-D subband [21] or 3-D wavelet packet [14] video coding. Therefore, wavelet packet decomposition currently attracts much attention for 2-D and 3-D video coding as well.…”

Parallel adaptive wavelet analysis

“…In particular, we treat the questions of data decomposition, boundary handling, and memory consumption. The algorithms are discussed in the context of 3D data as processed for 3D video coding [12] or the analysis of volumetric medical data [23]. …”