Fast reference frame selection based on content similarity for low complexity HEVC encoder

摘要: A fast reference frame selection algorithm for the HEVC encoder is proposed.A relationship between content similarity and derived.The studied without any extra computational complexity.Experimental results show that proposed efficiently removes encoding complexity of best decision process. The high efficiency video coding (HEVC) state-of-the-art standard, which achieves about 50% bit rate saving while maintaining same visual quality as compared to H.264/AVC. This achieved benefits from a set advanced tools, such multiple frames (MRF) based interframe prediction, improves encoder, it also increases heavy computation into encoder. becomes bottleneck definition videos be widely used in real-time low power multimedia applications. In this paper, we propose reducing prediction. Based large parent prediction unit (Inter_2N2N) children units (Inter_2NN, Inter_N2N, Inter_NN, Inter_2NnU, Inter_2NnD, Inter_nL2N, Inter_nR2N), information are obtained by learning their unit. Experimental can reduce 54.29% 43.46% MRF time low-delay-main random-access-main structures, respectively, distortion performance degradation negligible.