Fast block matching algorithm for H.264/SVC motion estimation based on sub-sampling
Hongsheng Yang, Jun Li, Jing Sun, Di Tian · 2010
In order to reduce the computational complexity of motion estimation based on block matching in H.264/SVC, a new fast block matching algorithm (BMA) is proposed. The algorithm consists of two parts. One is an adaptive sub-sampling BMA, which classifies BMA into three kinds: sum of absolute differences (SAD), half sub-sampling SAD, and quarter sub-sampling SAD, and predicts motion state of macroblock (MB) that is static or dynamic. The corresponding BMA is selected for MB according to the motion state of the MB. The other is an early termination algorithm for the calculation redundancy existing in BMA. Experiments of the different test sequences show that, compared with the fast search algorithm of the JSVM9.17 reference software, proposed algorithm can approximately reduce coding time by 40%, with similar PSNR and bit rate.