Directional wavelet transforms for prediction residuals in video coding

Abstract

Various directional transforms have been developed recently to improve image compression. In video compression, however, prediction residuals of image intensities, such as the motion compensation residual or the resolution enhancement residual, are transformed. The applicability of the directional transforms on prediction residuals have not been carefully investigated. In this paper, we brie y discuss differing characteristics of prediction residuals and images, and propose directional transforms speci cally designed for prediction residuals. We compare these transforms with the directional transforms proposed for images using prediction residuals. The results of the comparison indicate that our proposed directional transforms can provide better compression of prediction residuals than the directional transforms proposed for images.

Algorithms for Transform Selection in Multiple-Transform Video Compression

“…However, we note that the number of possible combinations is the number of transforms to the power of the number of blocks, which is very large even for the simplest case of two transforms. Some recent studies [2] have considered a set of as many as 17 different transforms that can be (a) A typical image chosen for each block. This implies that brute force method is not practical.…”

“…Recently, it has been observed that the statistics of motion-compensated residuals is different from that of typical images. Specifically, the work in [2] reports that the correlation of motion-compensated residuals can be better characterized as: By comparing the rate-distortion performance of three different settings, it is reported that for typical video sequences, much bit saving is due to the use of only vertical and horizontal 1D-DCTs.…”

“…In [2], the Lagrangian multiplier method is used. In this method, the rate profile associated with each transform is designed according to some proper choice of quantization and entropy coding.…”

“…Meanwhile, designing these codewords in terms of quantization and entropy coding is quite involved. In [2], the codewords are designed simply by modifying the existing quantization and entropy coding within H.264 system. Still, the rate-distortion performance increases significantly, indicating that these transforms work well on the residual images.…”

“…As discussed in the previous section, in the work of [2], a set of transforms that appear reasonable were first determined. For this given set of transforms, their rate profiles which depend on the quantization and entropy coding methods were obtained and incorporated into a practical H.264 system, which is a tedious process.…”