Research On Optimization Of Mode Decision Algorithm For H.264 In Real-time Monitoring System

Digital video monitoring system has become main stream among present monitoring systems. Among video coding standards used by digital video monitoring system, it is H.264 video coding standard that has the best coding quality. H.264 video coding standards is a new-type video compression standard jointly researched by VCEG of ITU-U and MPEG of ISO/IEC. Compared to previous video coding standard, H.264 video coding standard has made a large progress in coding efficiency and image quality. The progress in coding performance is due to that H.264 video coding standard has adopted advanced techniques, such as multi-block mode, multi-reference-frame, deblocking filter, and enhanced pixel motion estimation, different from previous video coding standards. When the introduction of the new techniques improves coding performances, it also complicates coding, facing H.264 with great challenges in real-time video coding and application in transforming. At present, how to decrease complexity of H.264 is one of hot problems in an area of video coding. The study shows that, applying numerous coding modes is one of the main reasons of the increase of coding calculation for H.264. Thus, this paper mainly researches the fast mode selection technology, which quickly determines macro block-type and guarantees image quality at the same time.Firstly, the development of video monitoring system and video coding standard is introduced in this paper briefly, then the new technologies of H.264 video coding standard in detail. Among the new technologies, it mainly researches the technology of multi-mode selection and its complexity, and then brings out the improvement in multi-mode selection algorithm. The previous fast mode selection algorithms have some deficiencies, such as no consideration of the characters of SKIP mode of inter-frame coding, no comprehensive consideration of inter-prediction mode and intra-prediction mode. In addition, the previous algorithms does not consider specific contexts. If we optimize fitting the features of video sequences in a specific context, optimization effect will be greatly raised.In compliance with the characteristics of video sequences captured by monitoring system, based on results from analysis of a sum of experimental data, this paper provides an adaptive fast mode selection algorithm varied according to QP. This algorithm contains optimization of inter-frame mode selection and prediction of intra-frame mode selection. In inter-frame mode selection algorithm, we firstly judge the SKIP mode. If the SKIP mode is the optimized coding mode of the current macro block, we will not consider other modes. Otherwise, we will use the complexity of Image texture to divide a macro block, so as to continue to choose other inter-frame modes. In choosing the inter-frame mode, we determine whether the inter-frame prediction should be terminated by the complexity of Image texture of the current macro block. In order to prove the above algorithm, we make experiments on the platform of the reference-modle JM8.6 of H.264. The experiment shows that in this algorithm, on high and low bit rate, coding quality is guaranteed, and the coding speed is greatly improved, especially it has better sequence optimization effect of monitoring videos with gently changed background.