The Study On Inter-Layel Intra-Frame Prediction For SVC To Extend MPEG-2 Base Layer

With the development of video coding technology, high-definition video has been gradually spreading all over the country, and the demand for fully high-definition becomes increasingly prominent. For digital TV, most families have been starting to use HD (high definition) STB (set top box) decoded by H.264/AVC, while some families are still using SD (standard definition) STB which only supports MPEG-2 decoding and can not play HD video decoded by H.264/AVC, which seriously hinder promotion of fully high definition. The existing solutions are to transfer HD video coded by H.264/AVC with same contents while transferring MPEG-2 SD video, which, although, solved the compatibility problem of STB HD and SD, causes plenty of redundant data due to transferring two ways of same contents but of different resolutions, causing a serious bandwidth waste.In order to solve the problems of STB compatibility and bandwidth waste, the article adopts spatial scalable coding thoughts in H.264 SVC Standards and provides the SVC based on hybrid standards, namely that with MPEG-2 and H.264/AVC as flexible base layer, it allows to regard MPEG-2 standard as base layer and H.264/AVC as enhancement layer to realize inter-layer prediction in standard room. The base layer adopts MPEG-2 coding standards, which can supports STB of older versions to ensure video playing; enhancement layer adopts H.264/AVC coding standards, which can enable new STB supporting HD decoding to play HD video. The proposal and design of SVC based on hybrid standards have perfectly solved the problems of STB playing compatibility and bandwidth waste. SVC standard inter-layer prediction includes three keywords, being inter-layer inter-frame prediction, inter-layer residuals prediction and inter-layer intra-frame predicdion. Since the base layer adopts MPEG-2 standards, there is no intra-frame prediction in this coding standard, and for Frame I, there is no residuals data for reference, it is impossible to conduct inter-layer intra-frame prediction for the enhancement layer. The article makes a study on the problem of inter-layer intra-frame prediction in SVC based on hybrid standards, proposing solutions in terms of compression efficiency and coding speed by analyzing Frame I image.In order to increase data compression efficiency, the article adopts the solution of strengthening Frame I inter-layer prediction. Since the base layer Frame I adopting MPEG-2 coding can not complete inter-layer intra-frame prediction, nor realize inter-layer residuals prediction, this article adopts Frame I image prediction, namely that predict sampling on re-constructed Frame I image of the base layer to enhancement layer, then compare Frame I image of enhancement layer with sampled Frame I re-constructed image to obtain residuals image. Since residuals image includes plenty of texture information and have excellent directivity, lastly conduct further itra-frame prediction to this residuals image to realize highly efficient data compression. The experiment shows that it can increase data compression efficiency and increase coding efficiency by enhancement treatment of Frame I.In order to increase coding speed, the article adopts intra-frame texture prediction solution based on DCT coefficient which can better reflect image textural properties and certain directivity. Therefore, one can conduct intra-frame texture prediction according to DCT coefficient, and realize rapid selection of intra-frame prediction mode according to DCT coefficient. The base layer of SVC based on various standards adopts MPEG-2 coding, and the article, based on the analysis on DCT coefficient in size of 8x8 of the base layer, adopts BP neural network model to conduct plenty of training to base layer DCT coefficient and intra-frame prediction model of its corresponding enhancement layer, and finally obtains DCT coefficient which can make conclusion on rapid selection of intra-frame prediction models to some extent, having increased coding speed and efficiency.The SVC based on hybrid standards studied in this article provides excellent theoretical basis to fully promote high-definition of digital TV, and further increases coding efficiency by Frame I enhancement prediction and design of rapid intra-frame prediction based on DCT coefficient, which has promoted coding technology development to some extent.