A hybrid block-matching approach to motion estimation with adaptive search area

An efficient block-matching (BM) method is proposed in this paper, which is a hybrid of full search (FS) and fast search block-matching algorithms (BMAs). The block-matching process is carried out in two stages. In the first stage, initial block-matching is applied to a few selected blocks in the frame using FS. To reduce computation, the search window size is adaptively adjusted based on the magnitude of the components of the motion vectors (MVs) in the previous frame. In the second stage, a pre-selected fast search algorithm is applied to the remaining blocks and the spatial correlation between the macro-blocks (MBs) is utilized to improve the searching performance. The proposed algorithm aims for fully exploiting the benefit of the optimal block-matching results generated by FS, and at the same time, remaining a low computation by adopting fast search algorithms. Experimental results demonstrate the effectiveness of the proposed method by improving the performance of the fast BMAs in terms of peak-signal-to noise-ratio (PSNR). Another benefit of the described approach is its independence of the fast algorithms, implying that most of the fast BMAs can be combined into the proposed framework for performance improvement.