An efficient adaptive energy model based predictive Motion Estimation algorithm for video coding

Motion estimation is a critical and time consuming part in most video encoders which affects highly the quality of output video sequence. Block matching techniques are usually used to estimate the motion. Among these, the Predictive Block Matching algorithms try to guess the location of the best matching block before searching for its decisive coordination. This paper proposes a novel Adaptive Energy model based predictive Motion Estimation (AEME) algorithm to measure a dynamic similarity criteria between blocks by comparing their energy histograms. Then an adaptive two step search algorithm is developed to estimate the motion of block. Simulation results show that AEME increases the accuracy of the prediction by 1.25 dB in contrast with its Median based counterpart. Additionally, the PSNR value presented by TSS algorithm is about 1.19 dB lower than AEME.