Inter frame

About: Inter frame is a research topic. Over the lifetime, 4154 publications have been published within this topic receiving 63549 citations.

Temporal domain sub-band video coding with motion compensation

Abstract: Temporal domain subband coding with motion compensation (MC-SBC) is a new technique of interframe data compression in video coding applications. Perfect reconstruction is possible with one-pixel accuracy for motion parameters and a trivial first-order quadrature mirror filter (QMF). With comparable complexity, even this most simple type of MC-SBC outperforms MC prediction (MC-DPCM) techniques when blockwise motion estimation is used. The concept of MC-SBC is generalized to subpixel accuracy of MC and any even-length (odd-order) QMF filters. Motion estimation is performed in a hierarchical forward-backward procedure which gives better SNR and visual performance results than blockwise-independent estimation. The scheme is extremely error resistant due to its nonrecursive structure; the gain over MC-DPCM is remarkably high, especially in layered coding schemes as they are discussed for ATM video applications. Results of MC-SBC and other interframe coding schemes are compared using two different intrafracture schemes in layered and nonlayered coding applications. >

A Flexible Structure for Fully Scalable Motion-Compensated 3-D DWT With Emphasis on the Impact

Abstract: We investigate the implications of the conventional " 2-D" motion-compensated (MC) three-dimensional (3-D) discrete wavelet/subband transform structure for spatial scal- ability and propose a novel flexible structure for fully scalable video compression. In this structure, any number of levels of "pretemporal" spatial wavelet decomposition are performed on the original full resolution frames, followed by MC temporal decomposition of the subbands within each spatial resolution level. Further levels of "posttemporal" spatial decomposition may be performed on the spatiotemporal subbands to provide additional levels of spatial scalability and energy compaction. This structure allows us to trade energy compaction against the potential for artifacts at reduced spatial resolutions. More importantly, the structure permits extensive study of the interaction between spatial aliasing, scalability and energy compaction. We show that where the motion model fails, the " 2-D" structure inevitably produces misaligned spatial aliasing artifacts in reduced resolution sequences. These artifacts can be removed by using pretemporal spatial decomposition. On the other hand, we also show that the " 2-D" structure necessarily maximizes compression efficiency. We propose different schemes to minimize the loss of compression efficiency associated with pretemporal spatial decomposition. Highly scalable encoders must operate with no knowledge of the actual decoded quality or resolution. The predictive feedback paradigm inherent in traditional motion-compensated (MC) video compression algorithms is incompatible with the requirement of highly scalable compression, as the encoder must duplicate the decoder's behavior during the encoding process to avoid the problem of error drift. Indeed, the preferred paradigm is that of feed-forward compression, in which a spatiotemporal transform is followed by embedded quantization and coding. At the same time, to efficiently exploit interframe redundancy, it is important that the transform estimates the motion between frames and compensates for it in the temporal decomposition. Three-dimensional (3-D) discrete wavelet/subband trans- forms (DWTs) are highly desirable for scalable video compres-

Bandwidth Reduction Via Movement Compensation on a Model of the Random Video Process

Abstract: The video-signal bandwidth reduction obtained by interframe differential transmission is hindered by rapid movements of the objects and the camera A simple movement compensation procedure can be obtained by dividing the image into zones and then transmitting for each zone a displacement vector relating the zone to a zone in the previous flame that is maximally correlated with it The effectiveness of such a procedure is studied on a model of the random video process, obtained as a derivation of a three-dimensional telegraph signal In particular, it is shown that for this model, the optimal procedure for finding the most correlated zone in the previous frame is the one matching the intersections of the contour lines The optimal size of the zone for which a single displacement vector has to be transmitted is also determined

System, method, and apparatus for error concealment in coded video signals

Abstract: Efficient techniques are provided to detect and correct channel errors found in an encoded video signal. In one example embodiment, header information in a video frame is analyzed to detect channel errors in the video frame. The video frame is then corrected for detected channel errors by isolating the detected channel errors to a few macroblocks in the video frame to reduce data loss and improve video quality.

Adaptive intra-interframe DPCM coder

Abstract: Adaptive prediction schemes provide lower transmission rates than those obtained by simple previous frame prediction. In this paper, we measure the entropy of prediction errors for two types of adaptive intra-interframe prediction algorithms. In the first case, that predictor which results in the least prediction error for previously transmitted neighboring pels is selected from a set of predictor functions. In the second case, prediction is a weighted sum of previous frame and intraframe predictions, where the weights are changed from pel to pel by gradient techniques. We also investigate various modifications of the basic methods. Further, a new type of variable length encoding in which the locations of the nonzero prediction errors are coded by horizontal run lengths is discussed. Compared with the pel entropy of previous frame prediction, the run length coding gives a gain of 2 to 16 percent, depending on the scene. Compared to simple previous frame prediction the first type of adaptive scheme in combination with horizontal run length coding provides a gain in entropy of 18 to 29 percent, whereas the second type of adaptive scheme provides a gain of 20 to 32 percent.