Fast Global Motion Estimation Based on Block Projection Match

Zhiguo Bian,Lizuo Jin,Shumin Fei
DOI: https://doi.org/10.1109/ccpr.2009.5344068
2009-01-01
Abstract:Global motion estimation is widely used in video process system, the algorithms such as KLT and SIFT is difficult to use in real-time system due to its huge computation. A new algorithm of global motion estimation based on block projection match is proposed in this paper. At first, the frame to be processed is divided into a few blocks. Then the motion vector of every block is estimated with the algorithm of projection match, and the corresponding weight of each block is assigned through an evaluation criteria. At last, the global motion parameter is acquired by an improved parameter estimation algorithm of RANSAC. The experimental results show that the proposed algorithm can satisfy the requirements of real-time applications without loss of accuracy.
What problem does this paper attempt to address?