Optimization Algorithm Of Inter Prediction For HEVC

With the development of computer multimedia technology, people's demands for service about HD or UHD video are largely increased, So it is a important goal to improve the ratio of video coding standard. JCT-VC has proposed HEVC(high efficiency video coding) standard.H.265/HEVC significantly improved the compression ratio of HD video,meanwhile, H.265/HEVC introduced a high degree of coding complexity.The inter prediction module has the largest time complexity.In terms of motion estimation, an optimization algorithm for integer-pixel motion estimation was proposed. The algorithm aims to reduce encoding time of inter-frame coding, while ensuring the same coding quality and compression ratio. Firstly, before motion estimating of a prediction unit, compare the motion similarities with coding unit at the previous division depth. Adopt the search template optimized if the similarity condition is met, else use the TZsearch template that provided by HEVC standard. Finally, save integer-pixel motion info for CU at next division depth. In terms of inter-frame mode decision, an optimization mode decision schemes was proposed, which includes a prediction algorithm of CU depth. This prediction algorithm could predict the approximate depth of current CU with the result of motion estimation.optimization mode decision schemes also use the correlation of PU modes to reduce complexity.Experimental results show that the proposed motion estimation algorithm reduces total encoding time by 14% on average, and bit rate increase not more than 0.3%, and the encoding quality is almost the same. On the basis of motion estimation algorithm, optimization mode decision schemes reduce total encoding time by 5% on average, no loss of coding quality.