Arithmetic Optimization Based On H.264

The rapid development of multimedia communication demands more videoprocessing and transmission techniques. Digital video services, which are representedby interactive real-time video communication, are the bottle neck of multimediacommunication because of their high technical requirements such as broad band, lowdelayandlowbit errorrate.Inthiscase,agoodstandardisneededespeciallywhenthenetworkbandwidthsfortransportaretakentoaccount.H.264/AVC is the newest international video coding standard. It was developedbyaJointVideoTeam(JVT)consistingofexperts from ITU-T'sVideoCodingExpertsGroup (VCEG) and ISO/IEC's Moving Picture Experts Group (MPEG). H.264 iscomposed of many advanced video coding techniques. Compared to existing videocoding standards, H.264 has made great progress in coding efficiency. For example, itimproves the compression efficiency by a factor of about 50% (on average) overMPEG-4. In addition, H.264 provides a very friendly network interface. However, theimprovement in compression efficiency of H.264 comes from the dramatic increase incomputational complexity; this makes it a must to reduce the computationalcomplexityofalgorithmsinH.264inrealtimeapplications.In this paper, the fundamental principles, methods and the standardization processof video compression is first briefly presented, then,the main functional modules ofH.264 decoder are introduced, including entropy decoding, inverse scanning, scalingand transformation, intra prediction, inter prediction, deblocking and reference picturemanagement. In addition, the main techniques on which H.264 is based to achieve itshighcompressionefficiencyispresentedindetail.Based on a thorough analysis of the existing CAVLC decoding algorithms, a fastCAVLC decodingalgorithm is proposedinthis paper. Basedon ananalysis ofexistingfast deblocking algorithms under MBAFF mode, a low complexity deblockingalgorithm is proposed. Extensive simulations on the methods proposed in this papershow, with little degradation in the decoded picture quality, the algorithms reducedramaticallythecomputationalload,andimprovedecodingefficiency.