Dct Domain Hdtv To Sdtv Transcoding, Image Sampling Algorithm And System Implementation And Optimization

Many international video compression standards have been established for various video applications, such as MPEG, H.26X. These standards differ in bit rate, frame rate and spatial resolution. As digital video application becomes more popular, there are increasing needs to dynamically convert the video between different formats. The transcoding of compressed video stream is to transcode a video stream from one format or standard to another. The transcoding technology can be divided into two types: one is to transcode the video stream from one format to another, which is called homogeneous transcoding; the other is heterogeneous transcoding, which transcode the compressed video from one standard to another.In this dissertation, the transcoding from HDTV to SDTV based on the background of DTV was discussed. First, the standards of MPEG-2 and H.264 were introduced and compared. Then two kinds of drift errors, which caused the decline of output video quality, were educed based on the theoretic analysis, and another kind of drift error, which was unique for transcoding in DCT domain, was studied. After the comprehensive consideration of different architecture, the DCT domain partial encoder transcoding scheme, which eliminated the drift error by motion compensation, was chosen as the system architecture. This architecture not only kept the video quality well, but also intelligently utilized the coding information of the input compressed video stream. Second, a limited rate algorithm for picture downsampling was improved, and the realization process was deduced in detail. The simulation results of different algorithms and theoretic deducing showed that this algorithm had relatively higher PSNR and lower computation cost. Third, the other key algorithms of transcoding in DCT domain was introduced, including: motion compensation in DCT domain, the choosing of macroblock pattern and rate control.In order to increase the efficiency of the transcoder, the system was optimized, the fast algorithms were proposed and MMX SSE2 instruction sets were utilized to increase the transcoding speed, after the analysis of the hotspots. Then the software realization was discussed and the performance of the system was tested. At last, a real-time demonstration system was implemented.