Hybrid Spatio-Temporal-SNR FGS Video Coding

With the high development of computer and network technology recent years, it is becoming true for people to enjoy VOD (video-on-demand) service by accessing network using PC. Streaming media technology changes the traditional multimedia service based on web. It allows clients to enjoy VOD service real-time, without waiting for whole multimedia file to be downloaded on local computer. Therefore, there is great commercial future for multimedia stream.However, to realize such VOD service, it demands great challenges for current video compression technology. Firstly, current network of Internet is heterogeneity. To cope with network's available bandwidth varying, bit-streams after encoded should provide scalable ability to adapt bandwidth's varying. Secondly, current best-effort Internet does not offer any quality of service (QoS) guarantees to streaming media over the Internet. Thus, bit-streams after encoded should provide robust error resilience. Finally, clients hope to obtain different services according to their available resource, such as different quality service, different frame-rate service and different resolution service. So, bit-streams after encoded should provide SNR, temporal, spatial and decoding complexity scalabilities.Scalable coding has already been recognized as an effective solution to resolve the network's bandwidth varying. Though traditional scalable video coding schemes provide scalability, the scalability is coarse and need expensive cost. To obtain fine granularity scalability, MPEG-4 standard adopts FGS coding scheme. FGS scheme permits bit-streams after encoded be truncated and transmitted arbitrarily, which matches bandwidth varying well. FGS scheme also provides robust packet-loss resilience and decoding complexity scalability. However, all the advantages of FGS scheme are obtained at sacrifice of coding efficiency. This is mainly due to lack of motion compensation on FGS enhancement-layer.To improve coding efficiency, a two-loop MC+FGS and a single-loop MC+FGS coding schemes are presented. The key idea of both MC+FGS coding schemes is using higher quality picture of enhancement-layer as reference. By improving the efficiency of motion compensation, both MC+FGS coding schemes obtain higher coding efficiency than FGS. In both presented MC+FGS coding schemes, an effective algorithm is presented to determine how many bit-planes...