Rate Distortion Optimization Based Screen Content Video Coding

To ease the pressure of storage space and transmission bandwidth,video coding has become one of the research hotspots in the academic and industrial circles at home and abroad.And the international image video groups have developed a series of video coding standards,continuously improving compression efficiency and widening the ap-plication scenes,including the classical video coding standards MPEG-1,MPEG-2,H.264 and so on.The compression efficiency of the newest generation of high effi-ciency video coding standard(HEVC)is about one times higher than that of H.264,arousing widespread concern and research in academia and industry.On the other hand,in the past few years,with the development of science and technology,the video ap-plication of screen content is getting deeper into our daily life,such as remote desktop sharing,cloud computing,second screen of mobile and so on.Compared with the tra-ditional camera-captured content videos,the picture characteristics of screen content videos have big differences.The traditional video coding methods are mainly designed for camera-captured content videos,thus when screen content videos are encoded with traditional video coding methods,the rate-distortion(R-D)performance degrades ob-viously.Therefore,it has strong theoretical and practical significance to study the key techniques in screen content video coding,to optimize the coding performance.Coding optimization with bit rate constraint and coding optimization targeting smooth qual-ity are two important applications of rate distortion optimization,which are also two important problems to be solved in screen content video coding.Therefore,aimed at screen content video coding with bit rate constraint,this dis-sertation proposes three rate control algorithms.Firstly,a fast pre-encoding based rate control algorithm facing off-line coding application is proposed,improving rate con-trol accuracy effectively;Then,a sliding window based rate control algorithm facing off-line coding application is proposed,reducing coding time effectively;Finally,a low delay rate control algorithm facing on-line application is proposed.Moreover,aimed at set-QP coding which targets smooth quality,this dissertation proposes an adaptive block-level QP setting method.The main contributions of this dissertation are:(1)Fast pre-encoding based rate control methodThis dissertation proposes a fast pre-encoding based rate control algorithm for non-uniform videos(including videos whose scene changes drastically and screen content videos).Based on the fast pre-encoding process,by studying the relationship between fast encoding and really complete encoding,we propose a global optimal bit alloca-tion scheme.Then we design a picture classification method,classifying the transition frame into scene change frame and demi-change frame.For scene change frames,the frame type is set as intra type,and an improved intra R-Q model is proposed corre-spondingly,for demi-change frame,R-? model parameter fixing method is proposed.This algorithm can outperform other methods in terms of rate distortion performance and rate control accuracy.(2)Sliding window based rate control algorithmThis dissertation proposes a sliding window based rate control algorithm for screen content videos,facing off-line coding applications.Firstly,a new bit allocation scheme for the first frame derived on the basis of R-D optimization theory is proposed.Sec-ondly,an adaptive length window split method is proposed,and the bits are allocated by complexity in the window.Finally,the pictures are classified into simple frames and complex frames by complexity,and independent inter R-Q models are built.This algorithm has good rate distortion performance and rate control accuracy,and the en-coding complexity is obviously lower than algorithm(1).(3)Low delay rate control algorithmThis dissertation proposes rate control algorithm considering low delay constraint,facing on-line coding applications of screen content videos.Firstly,we allocate bits for current frame according to picture complexity and buffer status,reducing the quality refinement as much as possible.Then,we proposes a linear R-MAD(rate-MAD,MAD:mean absolute difference)model to estimate frame QP.Finally,we propose a frame QP adjusting scheme to improve the accuracy of frame QP.This algorithm ensures good buffer status,the buffer hardly suffers overflow.Thus,this algorithm has good and smooth video quality and good bit control accuracy.(4)Adaptive block-level QP setting methodThis dissertation proposes an adaptive block-level QP setting method for set-QP encoding that targets smooth quality.Firstly,we study the quality refinement of static regions and translational regions in screen content video,specially the influence of qual-ity refinement on R-D performance.And then the influence of picture complexity on R-D performance is theoretically analyzed.Based on these analysis,the QP is set,thereby the video R-D performance will be optimized.