Research On Some Key Technologies Of HEVC

With the development of information technology and Internet, digital video technology iswidely used in communications, computer, radio, television and other fields. Video codingtechnologies has also been applied more and more.In video coding technologies, the performance ofH.264is the most prominent.Thus, H.264video coding has been applied widely as the mostinfluential standard of video coding. With the demand for higher quality video, H.264has beenunable to match the demand. Therefore, as the next generation video coding standard, HEVC videoencoding start to be studied, and it is expected to be launched in2012.The goal of HEVC is to provide better encoding method for audio and video services. HEVCvideo coding standard is based on H.264video coding standard. It holds some originaltechnologies of H.264, and improves some other related technologies. The specific research ofHEVC includes: improving compression efficiency,improving robustness, increasing error recoverycapabilities, reducing the time delay, reducing the channel access time,reducing random accessdelay, reducing complexity and so on. Therefore, in order to meet the higher needs of the videocoding, the study of HEVC video coding is important.The main contributions and innovation points of the thesis are as follows:1．The multiplicate scans and bi-directional reference fast intra prediction algorithm is proposed inthis paper. Firstly, this algorithm determines one or more scanning modes by the fringe relativity.Secondly, the final intra prediction modes are determined by preprocess and simple Rate DistortionOptimization (RDO) model. The simulation results show that compared with double scanning modeand bi-direction intra prediction algorithm, coding time is reduced greatly, bitrate is reduced andPeak Signal Noise Ratio (PSNR) is improved in the proposed algorithm.2．Rate control algorithm for adaptive prediction error and quantization matrix is proposed in thispaper. Firstly, this Algorithm determines whether Discrete Cosine Transform (DCT) is done for theMacroBlock(MB) according to the grads directional weight of macroblock and simple rate distortionmode. Secondly, Quantization Parameter (QP) is determined through a new binary target bitsallocation model,and the slope coefficient is calculated by the average slope weight in themacroblock. Finally the quantization matrix, which is determined according to QP and slopecoefficient, is applied in coding for the macroblock. Simulation results show that compared with ratecontrol algorithm of basic-unit in H.264, the proposed algorithm restricts coding time and bitratedramatically, moreover, PSNR is increased.3. Fast Adaptive Loop Filter Algorithm with The Optimal Symmetry (FALF_OS) is proposed in thispaper. Firstly, this algorithm determines the optimal symmetry filter according to area symmetry and average sum of absolute difference (SAD). Then the filter areas are obtained through theblock-based and quadtree-based method in I frame, and the filter areas are obtained through motionvector,motion estimate and RDO model in P or B frame.Finally the filter areas are filtered by theoptimal symmetry filter. Simulation results show that compared with Block-Based andQuadtree-Based Adaptive Loop Filter Algorithm (BQ_ALF), the proposed algorithm reduces thecoding time greatly,mean while, it keeps the reconstructed picture quality.4．Fast Multi-Symmetry Adaptive Interpolation Filter algorithm (FMS_AIF) is proposed in thispaper. The most optimal symmetry interpolation filter was determined by area symmetry propertyand simple RDO model. Simulation results show that compared with Adaptive Interpolation Filterswith Flexible Symmetry (SF_AIF), the coding time is reduced greatly, PSNR is increased farther andbit rate is reduced in this algorithm.5. A divide-sampling based hybrid temporal-spatial prediction coding algorithm is designed tofurther improve the coding performance of the conventional H.264coding. In the proposedalgorithm, a frame is first divided into four equal-sized subframes, and the first subframe is codedusing RDO model with inter or intra prediction adaptively. Then, the optimal prediction method ofthe MB in other subframes is selected flexibly and reasonably from intra prediction, the fast interprediction(FIP), and the spatial interpolation prediction(SIP). The simulation results show thatcompared with the conventional H.264coding, the proposed algorithms improved the codingperformance.All of the proposed algorithms in this thesis are simulated and examined, then compared with thecorrelative algorithms in order to prove their validities and advantages.