Research And Improvement Of Rate Control Algorithm Based On H.264

With the rapid development of computer network and wireless communication, the demand of multimedia service is increasing, video-coding technology is an important issue in video communication today, and it has a prosperous future in wireless communication, interactive multimedia and e-commerce etc. The research of video coding mainly includes two fields: the design of source model and the choice of coding mode. The former finds the statistics of the source and provides efficient coding tools in the coding standards while the latter chooses the best coding modes for different video sequences, which is deeply researched in the thesis.H.264 is the new standard developed in March in 2003 by ITU and MPEG. We first discuss the key technique in H.264 including system framework, Now H.264/AVC is the most leading video encoding standard and has broad application field. It combines all the merits of previous video encoding standard, grows on them and proposes a lot of advanced methods and ideas. The encoding efficiency and quality are improved very much.. its bit rate compare to forestall video compress standard reduce 50% averagely, with 700kbits/s code stream provide approaching the DVD quality, is the most one-up video coding standard in image communication domain. Therefore, get the attention increasingly, is the main direction for the video coding technique in the future years.In video code standard, standard code stream grammar semanteme, the decoder norm has done detailed regulation, but tactics and the algorithm have not made a request to the code. The coder function tremendous difference since code tactics and the algorithmic difference, rate control is one of the part being that these are most important in code tactics , is video code technology key component , direct effect to coded image mass and the creation exporting code rates. how to control the code parameter in order to engender bit flow accords with various application need and reaching higher code mass is that rate control the major objective studying. The code rate is controlled by changing quantization parameter size , springing to encode frame, code mode selected,rates , gain higher reconstruction image mass at the same time as far as possible. A perfect rate control method play an important effect to video code quality,The main research work of this paper mainly surrounds H.264/AVC video code standard and rate control technique to spread out.In this paper, the generation of h.264, applying prospect, system composition, the key technology have been analyzed at first. In order to achieve the higher coding efficiency,H.264/AVC adopt many new coding technologies, such as frame/field adapting select, inter-prediction, advanced intra-prediction,4x4 integer transform and quantity, advanced self-adapting entropy coding, loop deblock effect filter,error resume technique etc. Aim at different video content, H.264/AVC adopt a series of self-adaptive forecast and coding arithmetic, greatly enhance coding efficiency. At the same decoding quality, the coding quality of H.264/AVC enhance at least more than one double compare to MPEG-2 and H.263,at the same time the calculate complexity of the H.264/AVC encoder compare to other video compress standard, tower above at least several times even tens of times.Next, this paper discuss the principle of h.264 rate control,h.264 arithmetic adopt RDO to mode select, lead to use MAD forecast QP and come forth the Chicken and Egg Dilemma[45],it bring difficulty to the rate control of h.264.The existing two resolutions of h.264 rate control[45][22].one is JVT-F086[12] that improve MPEG-2 TM5[22] rate control, make, it to be the same with h.264;the another is JVT-G012[45],advance fluid flow traffic model to allot every basic unit coding bits, and use R-D function in the macroblock unit to calculate QP.JVT-G012 compare the two methods, consider JVT-G012 excel JVT-F086.This paper introduce the presently testing model adopting arithmetic-G012 rare control, it control separate in GOP layer, frame layer and basic unit layer, and use manifold techniques, including self-adapting basic unit layer, fluid flow traffic model, linear model for MAD prediction, the quadratic RDO model etc. adopt those techniques successful solve the cause and effectconflict between traditional rate control and h.264 RDO technique, can control the output code rate well and truly, the output video quality is also preferably, but meanwhile get the success also bring some questions. one is owing to use linear MAD model and quadratic RDO model, in the process of calculate MAD and forecast the two models'parameter, increase the complexity of algorithmic greatly.Finally, Through analysis of the rate control in H.264, a peak signal noise ratio-based (PSNR-based) complexity estimation method is proposed. The mean absolute difference-based (MAD-based) complexity estimation is adopted to predict the basic unit complexity, so that more accurate target bits allocation is achieved. With the computational complexity of the linear model for MAD prediction taken into consideration and using the space and temporal correlation among the sequences, an algorithm named the weighted model is presented to predict the MADs of the basic unit, which avoids the process of updating the parameters and reduces the computational complexity. Simulation results show that, the precision of the weighed model is comparable to that of the linear model. The proposed method can allocate the target bits and control the rate more accurately, and compared with JM8.6, reduce the PSNR fluctuations significantly and improve the average PSNR by 0.2dB.