Dynamic Motion Detection Techniques
A novel motion estimation technique with reduced complexity is introduced in this chapter. This is based on segmenting the current frame into active and inactive regions containing moving objects and stationary background, respectively. The inactive region is assumed with zero motion vectors. The early detection of blocks due to zero motion vector, leads to cut redundant computation significantly, which speeds up the coding of video sequences. New approach, based on frame variance and correlation, is proposed to decide the motion activity of a block. Block-based motion estimation has shortcomings like unreliable motion vectors, block artifacts, and poor motion-compensated prediction along the moving edges. A new search strategy based on edge detection is proposed to obtain more accurate motion prediction along moving edges.
KeywordsVideo Sequence Motion Vector Motion Estimation Motion Detection Current Frame
- 3.G. Bailo, M. Bariani, H.264 search window size algorithm for fast and efficient video coding with single pixel precision and no background estimation for motion detection, in Proceedings of IEEE CCNC2006, pp. 754–757, 2006Google Scholar
- 5.R. Korada, S. Krishna, Spatio-temporal correlation based ast motion estimation algorithm for MPEG-2, in Proceedings of 35th IEEE Asilomar Conference on Signals, Systems and computers, California, pp. 220–224, Nov 2001Google Scholar
- 6.J.Y. Nam, J.-S. Seo, J.-S. Kwak, M.-H. Lee, H.H. Yeong, New fast-search algorithm for block matching motion estimation using temporal and spatial correlation of motion vector. IEEE Trans. Consum. Electron. 46(11), 934–942 (2000)Google Scholar