Fast Block-match Motion Estimation Based on Multilevel Adaptive Diamond Search

AbstractIn this paper, a novel fast block-match algorithm called MADS based on multilevel adaptive diamond search is proposed. The algorithm adaptively estimates the frame-level motion complexity with the reference frame texture information and the macro-block residual value at first, and then estimates the block-level motion complexity according to the spatial-temporal correlation of the vector field. The threshold is applied to stop the stationary block from searching. The initial search point and different diamond search modes are adaptively selected based on motion type for non-stationary block. Experimental results show that MADS algorithm has better performance than other popular fast algorithms for a wide range of video sequences