什么是 BVOP ？【B-VOP - Bidirectional Video Object Plane】（英文版）

IsaacZ

B-VOP - Bidirectional Video Object Plane

By: Christine Martz



Meaning of B-VOP
B-VOP的意义

Bidirectional Video Object Plane is used for motion-compensated prediction processing in the creation of MPEG video streams.

MPEG-4 video streams can be divided into hierarchical layers. The lowest layer is the Video Object Plane (VOP) layer. It corresponds to a single frame of the video stream. VOP is used to represent rectangular-plane frames or arbitrary-shaped object plane.

Video scenes can be composed from several objects which may change in position, appearance, or size, all independent of each other. A past and a future reference Video Object Plane (VOP) is encoded in a way that for each object or shape motion vector, a predictor is built from neighboring motion vectors that were already encoded allowing for bidirectional prediction. B-VOP are bidirectionaly predicted from a past and a future reference Video Object Plane (VOP).




Other Related Definitions:
其他相关定义：

★ Efficient coding of digital video is achieved in accordance with this invention, by integrating the bidirectional prediction modes of the MPEG-1 and the H.263 standards into a single adaptive scheme, while eliminating the restrictions and limitations imposed in these standards. This results in an efficient yet flexible method for performing the bidirectionally predictive coding of pictures (improved B-pictures) that is capable of efficiently operating with good performance over a wider range of bitrates than that possible by equivalent techniques in the individual MPEG-1 and H.263 standards. The present invention is thus suitable for B-picture coding of the H.263+ standard. Furthermore, the inventive method can be applied to the bidirectionally predictive coding of either rectangular regions or arbitrary shaped objects/regions in video pictures (so-called B-VOPs) for MPEG-4. -[United States Patent]

★ A B-VOP is a VOP which is coded bidirectionally. For example, macroblocks in a B-VOP can be predicted using the forward, the backward or both using the forward and backward motion vectors; this has similarities to MPEG-1/2 in which B-pictures can use such motion vectors. However, MPEG-4 video also supports an H.263 based mode for motion compensation, referred to as the direct mode. In direct mode, the motion vector for a macroblock in a B-VOP is obtained by scaling of the P-VOP motion vector, and further correcting it by a small (delta) motion vector. -[AT&T Laboratories]

★ A VOP may be interpolated from data contained in previous VOPs and VOPs that have yet to be displayed. These would be called Bidirectional Interpolated VOPs or B-VOPs. A B-VOP would only be based on data contained in I-VOPs or P-VOPs and not on other B-VOPs. Motion estimation is necessary for encoding P-VOPs and B-VOPs and works by matching 16 x 16 or 8 x 8 pixel blocks with special attention being given to blocks that lie on the boundary of the VOP. -[David Mellor - Audio Media Magazine]


来源：http://www.birds-eye.net/definit ... _object_plane.shtml