Transcoding Based On Reducing Temporal Resolution

The key technique for the application of multi-media technology is to resolvethe conflict of the mass data of digital video/audio and the capability of digitalmedia and the bandwidth of communication network. The way is compressing. Forthis purpose, standards for video/audio encoding were made: MPEG-1 and MPEG-2.These two standards made digital video and digital television based on CD-ROMpossible. To satisfy the low bit-rate communication like video meeting and visibletelevision, more encoding standards were made: H.261, H.263 and MPEG-4.Nowadays, encoding standard of new generation is popular whose name is H.264.These encoding standards in the different fields were different from each other ongrammar, compression performance, output bit-rate and main encoding technique.But they have similar compression algorithm such as: inter-frame motioncompensation (MC), intra-frame DCT, quantification. These similarity make acondition for the transcoding of the support for different systems and the applicationof resource share cross network. The video transcoding means the aftertreatment forthe encoded video bit-stream to make it compatible for the receptor.Main works in this article:The framework, related concept, principle of video encoding and illuminatedall sectors of the video compression/encoding flow is introduced in this article. Dueto the strong temporal relativity of video sequence, motion estimation (ME) andmotion compensation could cut down the inter-frame redundancy of videoeffectively. Because of this situation, this article made an all-around introduction forthe concept and principle of motion estimation and motion compensation which arethe core techniques of video compression.This article made a systematic introduction for the popular standard H.264,including the overall structure of the encoder and the basic algorithm of eachmodule. Detailed descriptions were made for the new techniques of H.264 likereference frame technique, variable sizing block technique and intra-frame prediction technique.Several structures of the video transcoding were also introduced. Videotranscoding is mainly divided into video transcoding stream bit-rate transcoding,video transcoding stream resolution transcoding and video transcoding streamsyntax transcoding. To reduce the temporal resolution of video transcoding streamresolution transcoding, a new dynamic skipping frame mechanism with wideapplicability was proposed. Simple reuse of the input motion vector duringtranscoding may result in the descent of the video quality. So a new dynamic motionvector refinement based on dynamic skipping frame transcoding was proposed. Thatis to say the range of the motion vector was refined dynamically based on theaccumulated motion vector among frames. Simulation results showed that thismethod made significant improvement in the PSNR compared to the traditionalmotion vector re-estimation.