Design And Implementation Of High-efficiency Video Coding Algorithm Based On HEVC

HEVC is a new video coding standard released after H.264, and its main purpose is to meet the demand of increasing visual-quality of people. HEVC Encoder retains the traditional hybrid-coding framework, but compared with H.264, HEVC can save nearly 50% of the bit-rate in the same condition of video-quality. With the great improvement of coding performance, the complexity of the encoder is greatly increased. Therefore, how to minimun the complexity of video encoding while ensuring the video-quality is the main direction of HEVC research. After a thorough study of HEVC coding standard, combined with some existing optimization algorithms, some optimization algorithms applied in intra-prediction and inter-prediction are proposed.In order to reduce the complexity of the encoder, the paper firstly introduces the HEVC coding structure and key technologies involved, and analyzes the modules with high complexity in detail. Then, some new optimization algorithms are proposed. The main work is done as follows:1. Intra-prediction Algorithm. This paper firstly introduces the optimization method proposed by predecessors, analyzes their disadvantages and proposes a new method to obtain the main edge direction by the features of prediction unit.Secondly, after analyzing the relationship of the prediction mode between adjacent units, a fast algorithm that predict the modes of current PU with adjcent PUs is proposed. Finally, the depth of adjacent coding unit can be used to estimate the depth of current coding unit. Three fast algorithms above guarantee an increase of bit-rate within 1%and an decrese more than 10% in the encoding time.2. Inter-prediction Algorithm. The paper firstly analyzes the characteristics of inter-coding modes, gets the relationship between inter-coding modes and CU division,and designs a new CU early termination algorithm. The results show the algorithm can largely enhance the coding rate. Secondly, the paper analyzes the characteristics of the transform coefficients, and find out that the more the ratio of non-zero is, the more bits are used to encode them. Based on the statistics of video sequences, a new algorithm using the ration of transform coefficients is proposed to terminate the split of the transform unit. The result shows that the algorithm which can reduce the coding time and transmission rate largely enhance the efficiency of video coding.