An Adaptive H.265/HEVC Intra-frame Prediction Mode Selection Algorithm

In order to reduce the mass quantity data,many digital video compression coding standards have been proposed in the past 20 years.But with the increasing requirement of people to the quality of video quality,the classical standards may not meet the new demands.The latest coding standard — — H.265/HEVC emerges.However,the improvement of compression performance is with the higher computational complexity,which becomes the bottleneck of H.265/HEVC practical application.Therefore,it is quite important and necessary to reduce the complexity of H.265/HEVC coding.In this paper,an adaptive H.265/HEVC intra-frame prediction mode decision algorithm is proposed.Considering the correlation between SATD(Sum of Absolute Transformed Differences)cost in the rough mode decision(RMD)and RDO(Rate-Distortion)cost in the fine mode decision(FMD),we establish a s elective function controlled by a f actor which is context adaptive.The established selective function is used to measure the relative difference of two adjacent modes of SATD value,and it can choose prediction modes which have higher probability to be the optimal mode.According to the distribution of the SATD cost,the selective function can decide which modes can be discarded and which modes can be used for the FMD.In this way,the number of candidate modes for the FMD checking is reduced considerably,consequently to achieve the goal of optimizing the intra-frame prediction mode selection algorithm process.The comprehensive experiments on HM 13.0 show that the proposed algorithm can save 32% encoding time with only 0.9% BDBR increase.It implies that our proposed algorithm can reduce the computational complexity with negligible loss in coding performance.Compared with other works,our proposed algorithm achieves the state-of-art performance on the balance between performance loss and time saving.Therefore,our proposed algorithm can be used for all H.265/HEVC-based video encoder side to provide a technical basis for real-time realization.