Font Size: a A A

Research On Motion Estimation Algorithm And Optimization Of Real Time Video Code Systems

Posted on:2006-02-08Degree:MasterType:Thesis
Country:ChinaCandidate:X F KuangFull Text:PDF
GTID:2178360185496948Subject:Computer system architecture
Abstract/Summary:PDF Full Text Request
With the techniques of multimedia and network have been rapidly developed and widely applied in our society, it is becoming more and more convenient and quick for us to store, process and transmit digital multimedia signals. On the one hand, it is possible for the information makers to design their real time video codec systems based on common or dedicated processors under the capability development of those kinds of processors, on the other hand, the appearance of new video standards has showed that larger compression ratio demands more and more computation. In order to improve the performance of the coder, the MPEG-2 and MPEG-4 standards use motion compensation techniques based on fixed or variable block sizes, which brings the problem of increasing complexity and computation, and also, brings the challenge of implementing such real time video codec systems.In this paper, a new algorithm named coarse global motion estimation based block match algorithm using unsymmetrical-cross prediction (GMBUCVP) is proposed based on the features that the results of global motion estimation can coarsely represent the motion of the sequences. The GMBUCVP algorithm can keep good balance between the image quality and the required computation. The GMBUCVP first uses coarse global motion estimation to estimate motion parameters. Secondly, combined with more accurate motion vector prediction and early termination techniques, we take the advantage of unsymmetrical-cross search to fast approach global optimum and avoid searching the motion vector to a local optimum. Our experiment show that the GMBUCVP algorithm has some robustness in kinds of motion types.Also, in this paper, an implementation of new software architecture: macro block based MPEG-2 coder is proposed according to the hardware architecture of Digital Signal Processor (DSP). Comparing to the implementation of frame based MPEG-2 coder, our scheme is more applicable in this kinds of DSP like embedded systems. As for the optimization of global strategy and key component, we get the following result from our experience of optimizing our MPEG-2 coder: optimizing at the code-line level, is not nearly as effective as higher-order optimizations. Paging optimizations, cache line optimizations, and memory allocation optimizations can often have vastly more significant effects than code-line optimization. Algorithmic...
Keywords/Search Tags:Motion Estimation, Early Termination, PSNR, Pipeline and Optimization
PDF Full Text Request
Related items