As the name implies, object-based video coding is a concept of video compression that relies on the detection of arbitrary –shape objects in a video scene. The most widely known object-based video coding of scheme today is the ISO standard MPEG-4, which is an algorithm for coding of moving pictures and associated audio information. Unlike block-based video coders suck as MPEG-1, MPEG-2,H.261 and h.263,MPEG-4 detects entities in the video frame that the user can access and manipulate (crop, cut, paste, re-scale, move, rotate, etc. ), hence providing the user with content-based functionalities for the processing and compression of any video scene. These entities are called Video Object Planes ( VOPs ). Therefore, in addition to coding the texture and motion information traditionally encountered in block-based video coders, MPEG-4 codes the shape of each VOP, so that the composition of objects could be done at the decoder end.
A VOP can be a semantic object in the video scene; it consists of luminance and chrominance components (Y, U, V ) in addition to shape information. The VOP can be of arbitrary shape and can be detected by semi-automatic segmentation. Each detected object in the frame is a VOP with different contour, texture and perhaps motion specifications.
When the video sequence has only one VOP of rectangular shape that is displayed at fixed intervals, it corresponds to the frame-based coding technique employed by block-based video compression algorithms. In order to code VOPs whose width and height are not integer multiples of 16 (MB size), the width and height of these VOPs are then padded using a repetitive padding technique described in Section 2.5.4.MPEG-4 supports two ranges of output bit rates. The first is tailored to low bit rate video communications below 64 k bit/s for low-detail video scenes, and the second is focused on high bit rate video applications (mobile TV for instance ) up to 2 M bit/s for height-activity fine-detail sequences. In addition to that, MPEG-4 defines a suite of error resilience algorithms to improve the robustness of coded bit streams to channel errors. Error handling mechanisms employed in video codes are examined in Chapter 4.
2.5.1 VOP encoder
The VOP encoder structure is composed of two parts: the shape coder and the traditional motion and texture coder that is found in block-based video coders such as H.263.
2.5.2 Shape coding
Two methods are employed by MPEG-4 to code..