Research On Fast Mode Decision Algorithm In Scalable Video Coding

Video signal is encoded by the way of single layer in the previous video coding standard, even though the coding efficiency is very high, but the network transmission condition and the computable capacity of terminal devices is not considered, so once the coding is finished, the bit stream is fixed. If the broadband of network fluctuating or the terminal devices' computable capacity is not enough, the transmission will fail or the bit stream that has been received can not be decoded.So,we need a new video coding technology whose bit stream is full of scalability, and which deal with the challenge of transmission of cross-network. The JVT (Joint Video Team, JVT) present a new video coding standard with the scalability. The emerging SVC chooses the Extension of H.264 as the framework. The Extension of H.264 incorporates efficiently the function of scalability of temporal, spatial and quality as well as contains the good encoding performance of H.264. The extension of the H.264,inheriting all the sophisticated coding tools from the H.264/AVC, reduces markedly the gap between the single-layer video coding and scalable video coding, also the bit stream encoded by the SVC have the ability of scalability of temporal, spatial and quality or their combination and can provide multiple subset bit stream which is decodable. This scalable bit stream can be extracted with the change of broadband of network and the capability of the terminal devices.In SVC ,in order to acquire a high coding performance, not only the operation of H.264/AVC,but also various tools such as the cross-layer prediction is needed to performan,so those technologies results in extremely large encoding time which obstructs it from practical use. So, currently, reducing the complexity of the SVC becomes gradually the hot topic.Firstly, this paper reviews the base framework and all kinds of scalability tools of SVC is introduced, especially the cross-layer prediction in spatial scalability is present, the experimental results of every scalability is given respectively.This paper then studies intra-mode decision strategy in H.264/AVC and the enhancement layer in SVC, based on the mode spatial correlation and mode inter-layer correlation, in addition to the feature of the SKIP mode, combining a fast selective intra-mode search algorithm based on adaptive threshold scheme in H.264/AVC,proposes a fast intra-mode decision algorithm based on adaptive threshold scheme in SVC.This algorithm can cause at least 80% marocblocks skip the intra-mode prediction process, and in the single coding for the IPPPP sequence type, the overall encoding time can be reduced up to 27%for the IPPPP sequence type , with little loss of image quality.This paper next analyze one of fast inter-mode decision process, realize the fast mode decision algorithm based on temporal, and making use of inter-layer motion prediction mechanism, improve the fast mode decision procedures based on spatial, and implement this algorithm in JSVM 9.8. Experimental results are given respectively.At the last part of this paper, base on the feature prone to transmission of SVC, a demonstration system of simulator for SVC transmission and decoding is realized.