Two fast multi-point search algorithms for high quality motion estimation in high resolution videos
Marcelo Porto, Diego Noble, Luciano Volcan Agostini, Sérgio Bampi · International Journal of Information Technology Communications and Convergence · 2011
In this paper, we present two new algorithms focusing on a high quality fast motion estimation process for high definition video coding. Both algorithms provide more efficiency to avoid falling into local minima in fast motion estimation when compared to diamond search (DS) algorithm. They were called multi-point diamond search (MPDS) and dynamic multi-point diamond search (DMPDS). The multi-point search could be done in a serial or parallel approach. In the parallel approach, the penalties in the performance are minimal near the impressive gain in final quality. The MPDS and DMPDS algorithms were implemented and evaluated on ten HD 1080p video sequences. The results show an average quality gain, in comparison with original DS, about 3.9 dB and 4.55 dB for MPDS and DMPDS respectively.