Chds: a Fast Search Algorithm for Motion Estimation in Video Coding Standards

In this paper, we propose a new hybrid Cross Hexagon Diamond Search algorithm (CHDS) using crossshaped search pattern as the initial step and asymmetric hexagon-shaped patterns and small diamond pattern as the subsequent steps for fast block motion estimation. In block motion estimation, search pattern with different shape or size and the center biased motion vector distribution characteristics has a great impact on search speed and distortion performance. The previously developed fast search algorithms focus on improvement of either coarse search or inner search. The Proposed CHDS algorithm reduces the number of search points by exploiting the distortion information in the neighbouring search points. Simulations are done using MATLAB and our experimental results indicate that the proposed CHDS algorithm reduces an average of 39.78% of time for motion estimation compared to New Hexagon Search (NHEXS), 10.86% of time for motion estimation compared to Hexagon Search (HS) and 38.64% of time for motion estimation compared to Diamond Search (DS).