Fast intra coding based on online learning for high efficiency video coding

The state-of-art high efficiency video coding (HEVC) standard attracts much attention as its excellent performances in video compression in recent years. However the inherent coding complexity of HEVC makes it difficult for real-time application. A fast algorithm for HEVC intra coding based on online learning is proposed in this paper. During learning coding, the partition threshold of coding unit (CU) is obtained by learning the random sample set which conforms to Poisson probability distribution. Meanwhile the threshold of rate distortion cost is also obtained by online learning. After that, the acquired thresholds are used for early skip of CU depth or early termination of CU partition in the stage of fast coding. Anchoring the HM16 test model, the proposed algorithm achieves an important improvement of average time-saving 46.02%. In addition, it is also demonstrated that the proposed algorithm gains higher time efficiency when video sequence is encoded with higher compression efficiency.