| 摘要 |
<p>ISO/IEC have standardized a video bit stream representing a compression method for standard resolution as well as high resolution video signals and pictures, the so-called MPEG-2 standard. MPEG-2 provides for three different picture coding types: Intra, Predicted and Bidirectionally predicted or interpolated frames or fields. In P and B frames or fields macroblocks can be coded intraframe or intrafield, respectively, if motion prediction does not give good results. The decision has to be made in the encoder with criteria to be determined by the implementation. In addition, MPEG-2 allows for frame or field based DCT which requires sorting of lines and another decision to be made by the encoder. A real-time MPEG-2 video encoder has been developed based on a set of dedicated IC designs. The processing functions are organized in three main ICs or stages. A preprocessor contains all input and pre-processing functions like picture format conversion which are required for data compression and a coarse and fine motion estimator contains all motion estimation and motion prediction functions. The bit stream encoder includes all bitstream generation for MPEG-2 encoding like DCT (420), quantizer (430), inverse quantizer (431), inverse DCT (421), zigzag scanning (412), and RLC/VLC encoding (422), but does not include motion prediction. This bitstream encoder uses pipeline processing and contains a multi-purpose bit stream assembling stage (432). <IMAGE></p> |
