Research On Rate Control Algorithm In Video Coding

With the development of multimedia communication and networktechnology, people's requirement of video application is more extensive.Video information plays an important role in people's life and socialdevelopment. Rate control technology is an essential component in videocommunication application system and it adjusts encoding parameters toobtain the best video compression quality under the constraint of channelbandwidth and storage capacity. An effective rate control scheme can ensurethe video quality and improve the utilization rate of the channel at the sametime, thus it has important significance for breaking through the limitation ofbandwidth and reduction of storage cost.As a new generation of video coding standard H.264/AVC, due tointroduction of many advanced coding technology, it has made significantimprovement in coding efficiency, compression quality and networkadaptability compared with previous standards. However, due to introducingthe rate-distortion optimization (RDO) for mode selection, quantizationparameter (QP) is used both in rate control an RDO, which makes ratecontrol of H.264/AVC more complicated and results in a so-called "chickenand egg" dilemma. In a given channel bandwidth, how to effectively controlthe virtual buffer fullness and obtain better visual quality, while reducing thecomputational complexity of the rate control algorithm has become the hotspot of research in this area.In this paper, we systematically introduce the basic theories andworking principles of rate control in video coding. Based on summarizingand analyzing the proposal drafts and its improved algorithm of rate controlfor H.264/AVC, then we studied the GOP (Group of Pictures) layer, frame layer and macro-block layer rate control algorithm in deeply. Our researchesinclude the following parts:Scene change occurs frequently in actual video, which causes negativeimpact on the rate control. In this paper, we focus on scene change of videosequence and then research scene change detection as well as its processingmethods. Finally, an adaptive variable length GOP layer rate controlalgorithm is proposed. In the proposed algorithm, the length of current GOPchanges adaptively with the ocurrence of scene change, so that the entiresequence contains a fixed number of I frame to avoid the actual bit rateincreasing. Furthermore, it adjusts the QP of the scene change frame to easethe pressure of the buffer. Simulation results show that the proposedalgorithm improves the video quality significantly on the basis of meetingbit rate constraints accurately, which shows better adaptability for sequenceswith scene change occurrence frequently.Initial quantization parameter affects the performance of rate controlgreatly. Selecting an appropriate initial QP not only can significantlyimprove the objective quality of video and precision of rate control, but alsocan reduce the fluctuation of image quality and then improve the subjectivequality of the video. In this paper, by using gradient as a complexity measure,an initial QP calculation method is proposed based on the temporal andspatial characteristics. By adopting the new method, the calculated initial QPis more consistent with the real property of sequence. The experimentalresults show that the proposed algorithm achieves better rate distortionperformance, on the basis of meeting bit rate constraints accurately.In this paper, based on the discussion of I frame QP calculation methodof rate control and its inadequacy in buffer control, coding quality as well asadaptability to different initial QPs, an adaptive I frame rate controlalgorithm is proposed. The proposed scheme adjusts the QP of intra-frameadaptively according to intra-frame buffer fullness ratio, current bufferfullness ratio, and skipped frame ratio. Experimental results demonstrate thatthe proposed scheme can obtain accurate bitrates and better PSNR (Peak Signal to Noise Ratio) performance while significantly reducing the numberof skipped frames, which effectively improved the subjective and objectivequality of the video. More importantly, the proposed scheme is very flexiblefor different initial QPs and target bitrates.In this paper, based on the analysis of temporal-spatial correlation andtexture direction between macroblocks, a temporal-spatial adaptivemacroblock layer rate control algorithm is presented,which including MAD(Mean Absolute Difference) prediction model, reference macro blockselection scheme for parameters of quadratic R-Q (Rate-Quantization)model updating and head bits prediction model. The proposed algorithm cannot only obtain more accurate predictive value of MAD and head bits as theinput of quadratic R-Q model, but also can get more accurate quadratic R-Qmodel coefficients. The experimental results show that the proposedalgorithm improves the quality of the video, on the basis of meeting bit rateconstraints accurately.