Implementation And Optimization Of An Video Encode System Based On DSP

With the development of information and communication technology, voice communication is popular in modern society. However, as a more direct information exchange tool, video communication has not been applied widely, for its tremendous amount of data and complicated processing mode. But the great application prospect and commercial opportunity of video communication technology have make the research and development of digital video information's storage, transmission, search, analysis as hot topic. On the one hand, general video data has been compressed under several hundred Kbps through innovation of video compress theory, improvement of video compress algorithm and modification of video compress standard. On the other hand, the techniques of electron, communication and computer being developed at very fast speed, widespread real time video communication will be come true before long.In this paper, we introduce features and main algorithms of MPEG-4, which became an International Standard for coding of audiovisual information in December 1998. Texas Instruments (TI) TMS320C62x digital signal processor (DSP) is analyzed, which Allow run at a 600 MHz clock rate and 4800MIPS and achieved the higher performance than earlier digital signal processing solutions. Subsequently, MPEG-4 video compress encode is implemented on the DM642 DSP through improvement of encoding algorithm of MPEG-4 and optimization of run rate of program. At the same time, capture, transfer, play and other functional modules in the video encoding system are introduced.There are some optimization steps executed including reforming of algorithm for running DSP rapidly, using of specific instruction, expanding of soft turnover for improving stream line rate, dispatch of space of memorizer to increase hit ratio of cache, introducing of many EDMA transfers to reduce load of DSP kernel.It is probably for three-dimension wavelet algorithm become research trend of compress algorithm of next period. Three-dimension wavelet is also investigated and discussed in this paper. Adapting grouping is introduced, and distortion resulted from acute motion is improved, attributed to using of 3D SPTHI video compress method based on three-dimension wavelet.