Fast Block-matching Motion Estimation by Recent-biased Search for Multiple Reference Frames

Multi-framc motion compensation improves the rate-distortion performancc substantially but introduces much higher loading to the system. Without considering temporal correlations, convcntional singlc-frame block-matching algorithms can bc used to search multiple frames in a rather inefficient frame-byframc way. In order to exploit the motion characteristic in longterm memory, a multi-frame extension of the wcll-known crossdiamond scarch algorithm is proposcd. Unlike thosc algorithms that evenly scarch each reference frame, our algorithm adopts a novel recent-biased spiral-cross search pattem to sub-sample the 3-dimensional mcmory space as a whole. This approach significantly boosts the efficiency ofthe block-matching process. Two new tcchniqucs, starionmy block tracking and multiple searchingpaths, are employed to further improve the spced and accuracy. As compared to full search, experimental results show that our algorithm can reduce up to 99.5% complcxity in terms of scarching points while limiting the PSNR loss in 0.04dB. Simulations also prove that our algorithm out-performs the cross-diamond search and diamond search algorithms in speed and accuracy.