Fast predictive motion estimation algorithm with adaptive search mode based on motion type classification

In this paper, we propose a fast predictive search algorithm (FPS) that is based on diamond search (DS) algorithm but adding initial searching point prediction, motion type classification, and early searching termination techniques. In the proposed FPS, the start search point is set near the optimized search point in the previous search step, which can avoid inefficient global search procedure and can find the initial search point efficiently. Based on motion type classification, the adaptive search pattern for both maintaining image quality performance and reducing search points is provided in FPS. By combining the stationary state judgment and motion type classification prediction together and adopting the early searching termination criteria with different fixed stationary threshold settings for different motion types, FPS can lower the search points in a precise, efficient, and practical way. The comparison results show that the proposed FPS motion estimation algorithm can reduce 60% search points, while image quality is similar as DS algorithm.