HEVC Complexity Optimization Algorithm Based On Temporal And Spatial Correlation

In recent years, with the HD video,Ultra HD video(resolution to 4kx2 k,8kx4k) application into people's horizons gradually, video compression technology has been a huge challenge. In addition, a variety of hardware and software is constantly updated to make a wide variety of video applications continue to emerge.A more efficient video compression performance is also presented in the demand for diverse video applications. To this end, VCEG and MPEG in April 2010 once again set up a joint video coding group, jointly developed a new generation of video coding standard- H.265/HEVC. Compared with the previous coding standards, HEVC coding standard for the transmission of the same quality can save nearly 40%-50% of the storage space and bandwidth, and under the condition of the same bit rate can be increased by nearly double the video quality. At the same time, HEVC coding also brings high complexity, which makes its wide application and development restrict. The main research work include: Firstly,the overall process of HEVC coding is introduced, and then the key techniques of HEVC are introduced in detail. Then, the CU mode selection process is introduced in detail, and studying and discussing the existing CU block partitioning optimization algorithm. Next, through the data analysis, verifying the strong correlation between adjacent CU units. Based on this, this paper presents the encoding complexity of optimization algorithm, the algorithm is constituted by the two aspects: based on the relevance of the encoding depth to predict the current CU depth and choose the skip prediction mode early. In the division of the coding unit depth: first of all, we analyze the strong correlations between the CU and the adjacent CU, including current CU left, above and to the left of the CU and the previous frame the current CU corresponding position of CU. Base on this, this paper considers a preliminary selection for the depth range of the current CU base on the optimal depth partition of the encoded units near the current CU, only to calculate to those most likely depth value. Then, when traverse the prediction mode of the current depth, if the current CU texture is simple, then analyze the situation that the SKIP model is selected as the best prediction mode by the adjacent CU, and combined with the space-temporal correlation of coding unit, when the maximum coding unit skip the depth partition, judging the best prediction mode of the surrounding CU, if match condition, then select SKIP mode as the best prediction mode of the current CU, terminate the calculation of the rate distortion of other prediction modes early.Experiments show that the optimized algorithm reduced the computational complexity, shorten the encoding time.