Research On Technologies For Next Generation Video Coding

International standard organizations have devoted to establishing the next generation video coding standards after the H.264/AVC video coding standard was published. The establishing video coding standards includes high performance video coding (HEVC) standard and 3D video coding (3DVC) standard. This thesis studies some technologies about HEVC and 3DVC. The main contributions of this thesis are high efficiency intra frame coding, high efficiency motion vector prediction for videos with zoom motion, the virtual view's distortion model in 3D video coding, and the model based joint bit allocation algorithm between texture videos and depth maps for 3D video coding. The principal contributions of this thesis are concerned with the following aspects:1. Decoder side prediction mode derivable intra coding methodFor a certain block, since different intra prediction mode could give similar predicted blocks; it is no use to signal the prediction mode in the bit stream for the block. All the predicted blocks are combined together to obtain the final predicted block. The variance of the mean value and that of the variance value of the pixel values in each predicted blocks are used to determine whether all the predicted blocks are similar or not. RDO process can be neglected in the proposed method, and the identifiers of prediction methods can also be omitted as well. Experimental results show that compared with H.264/AVC, the average bit rate saving and coding time saving are 2.62% and 39.25% respectively when the quality of reconstructed videos are the same.2. Model based motion vector predictor for zoom motionAs zoom motion is common in video applications, a linear motion model is derived to describe zoom motion based on the analyses of camera imaging principles. Based on the motion model, a motion vector predictor for videos involving zoom motion is proposed. A rate distortion (RD) criterion is used to choose the optimal motion vector predictor between the one utilized in H.264/AVC and the one derived from the linear motion model. Experimental results demonstrate that by implementing the proposed method into Key Technology Area test platform version 2.2r1(KTA2.2r1), the maximum and average bit rate savings can be achieved as high as 7.66% and 4.90% respectively, while maintaining the same quality of reconstructed videos.3. Distortion model of synthesized virtual views for 3D video codingThe distortion of the virtual view which is synthesized from its neighboring texture videos and depth maps is analyzed in detail. The distortion of the synthesized virtual view is decomposed into three parts, one is induced by the compression of the neighboring texture videos, the second one is induced by the compression of the neighboring depth maps, and the third one is the inherent geometry distortion which is induced by accuracy of depth maps. Thereafter, the relationship among the distortion of the synthesized virtual view, the compression distortion of the neighboring texture videos and that of the neighboring depth maps is derived. Experimental results demonstrate the accuracy of the derived distortion model.4. Joint bit allocation algorithm for texture videos and depth mapsThe joint bit allocation algorithm for texture videos and depth maps is designed based on the derived distortion model of the synthesized virtual view. Based the derived model, the bit allocation problem was formulated as a constrained optimization problem, and was solved by Lagrangian Multiplier Method. Since the optimal QPs of texture videos and depth maps can be calculated numerically, it is no need to search the optimal QPs. Full search algorithm; Liu's algorithm and fixed ratio 5:1 bit allocation method were used as the benchmarks to demonstrate the performance of the proposed algorithm. Experimental results show that the distortion model is accurate, and the R-D performance of the proposed algorithm is close to those search based algorithms, i.e. full search algorithm, Liu's algorithm, while the complexity of the proposed algorithm is low. Besides, compared with the fixed ratio 5:1 bits allocation method, a maximum 1.2dB gain can be achieved by the proposed algorithm.