Motion estimation and compensation based on almost shift-invariant wavelet transform for image sequence coding
✍ Scribed by Mohammed A. Al-Mohimeed; C. C. Li
- Publisher
- John Wiley and Sons
- Year
- 1998
- Tongue
- English
- Weight
- 575 KB
- Volume
- 9
- Category
- Article
- ISSN
- 0899-9457
No coin nor oath required. For personal study only.
✦ Synopsis
There has been rapid progress in the application of referred to as intraframe coding, whereas those which deal with wavelet transforms to image and image sequence compression. The the temporal correlation are called interframe coding. In videostandard discrete wavelet transform lacks translation invariance in coding systems using wavelet decomposition or subband decomimage decomposition which will affect the accuracy of motion estimaposition, there are several schemes used to remove the redundantion from the decomposed subimages in video coding. In this article, cies. The main difference between these schemes is in the order we present a study of applying an almost shift-invariant wavelet transin which the spatial and time redundancies are removed. The two form with ''oversampled frames'' to image sequence compression.
most common approaches of Hybrid video coding are discussed
With minimal oversampling and biorthogonal spline wavelets in the in the following paragraphs.
almost shift-invariant wavelet transform, motion vectors can be more Scheme I, as used by Motion Pictures Experts Group (MPEG) accurately estimated, contributing toward fewer prediction errors in comparison to those obtained with the standard discrete wavelet (MPEG-1 and MPEG-2) and H.261 standards, removes first the transform. Thus, an improved compression ratio can be obtained. We temporal correlation with frame prediction; then, the spatial correpresent two new algorithms, the full-motion oversampling algorithm lation is applied on the prediction error. These standards do not (FMOS) and the reduced search multiresolution motion estimation use subbands or wavelets, but rather, rely on macroblock structure algorithm (MRME), for estimating motion fields at different scales to cope with the motion estimation. The discrete cosine transform and in different subimages. In the latter, motion vectors at a higher (DCT) is used in these standards to remove the spatial redunresolution are approximated by the motion vector estimates at a lower dancy. Codec using macroblocks in spatial correlation, such as resolution through proper scaling. Experiments were performed on DCT, or temporal correlation suffers from blocking artifacts in three video sequences with a variety of motions including slow, fast, extreme cases. and zooming. Our results have shown that both algorithms, FMOS Ebrahimi [1] proposed a codec similar to scheme I, but using and MRME, using the almost shift-invariant oversampled frame wavelet transform have reduced prediction errors and enhanced the com-Gabor-like biorthogonal and linear phase wavelets as a variant pression performance in terms of peak-signal-to-noise ratio (PSNR) of the above standards. In contrast to the MPEG algorithm, Ebrafor the same bit rate when compared to the existing full motion himi's codec operates on the entire frame and no macro block standard algorithm.