Research of Simulation on Global Motion Estimation Based on Successive Elimination Weight Algorithm and vectors accumulation
Maoliu Lin · Jisuanji fangzhen · 2007
A global motion estimation based on the Successive Elimination Average algorithm and vectors accumulation is proposed in the paper. The reference frames to be estimated are analyzed first, the vectors accumulation algorithm is proposed based on the hypothesis that the error plane is monotone around the global match point in a small scale. And to satisfy the goal of the global motion estimation, the directions of local motion vectors and the distances between local motion vectors and the average are judged firstly after the local motion vectors estimated. The local ones that cannot meet the demand requested by the algorithm are eliminated and the left vectors are given different weight while being calculated in the later global motion estimation. Experimental results with several image sequences show that the performance of the proposed algorithm is very good; PSNR has increased 5.77%～34.18%.