Adaptive-centre diamond search distance-dependent thresholding search for fast and performance-scalable motion estimation
Sorwar, G 2004, 'Adaptive-centre diamond search distance-dependent thresholding search for fast and performance-scalable motion estimation', in MH Hamza (ed.), Proceedings of 6th IASTED International Conference on Signal and Image Processing, Honolulu, Hawaii, 23-25 August, IASTED, pp. 453-458.
Fast and performance scalable motion estimation algorithms for compression of moving pictures are widely required for real time video encoding applications. Most of the existing fast algorithms such as three step search (TSS) and new three step search (NTSS) are application dependent and do not provide any flexibility in controlling the quality of service (QoS) in terms of predicted image quality and computational complexity. The authors have addressed these issues by proposing a novel motion estimation algorithm by introducing the concept of a distance dependent thresholding search (DTS). Though, the performance of DTS algorithm was well for low motion video sequences, it was not so satisfactory for high motion video sequences. In this paper, DTS algorithm has been modified into adaptive-centre diamond search distance-dependent thresholding search (ADSDTS) algorithm by incorporating a diamond-search pattern instead of square search pattern in DTS algorithm, and adaptively predicting the initial search center for each current block, for trading off quality and processing speed. Experimental results show that the proposed ADSDTS algorithm that provides the full flexibility for performance management in motion estimation for video coding, also outperforms the fast TSS, NTSS, and adaptive centre NTSS (ACNTSS) algorithms by reducing the computational time with all motion types video sequences.