Research On Scalable Video Coding Based On MPEG-4

Video services are expected to be one of the important applications supportedby the next generation networks. The video traffic is high bursty so that efficientvideo traffic management made network more complicated such as ATM networkdoes. But simple protocol can't provide guaranteed QoS and efficient networkutilization simultaneously. So cost efficient video stream processing techniques forimproving network utilization with guaranteed QoS is very important and worthy.In the paper, common video compress standard and compress method are firstlyintroduced, and then the latest video compress standard MPEG-4 is detailedlyanalyzed, on the base, an encoder compatible with MPEG-4 was realized on theVisual C++ 6.0 development station.In the paper, some used flexible coding was analyzed, based on the merit andshortcoming, a kind of coding method is put forward which is combinative withspatial scalable and fine gradual scalable, it is named as Fine Granular SpatiallyScalable-FGSS.As the experiment result shows, traditional spatial scalable codingonly provide two kinds of video quality level, it is only narrow scale that the spatialscalable coding is better than FGSS in coding efficiency when the enhanced layer iscompletely decoded, and the coding rate of FGSS is chose in a widest range, at thesame time, video quality increased with the bit rate;In a low bit rate, FGSS is moreefficient than PFGS, while in a high bit rate, FGSS is the same as PFGS;That is tosay that FGSS is more adaptable for network.Furthermore, in the paper, a adaptive arithmetic is put forward which canautomaticly choose a macro block model among 4 models according as it is theminimum DCT coefficient, it is to improve image quality of video frequency and toadvance the compression. And the relation of motion vector in different quality levelis analyzed, it is beneficial for error resistibility.