A kind of global motion estimation algorithm based on feature matching
Shuxiang Guo, Chenguang Qiu, Xiufen Ye · 2009
In this paper, we propose a kind of global motion estimation algorithm based on feature matching. The scale invariant feature transform (SIFT) algorithm is applied to global motion estimation. The feature extracted by SIFT algorithm is invariant to image scale and rotation. The matching accuracy is very high even under the condition of additive noise, varying illumination and affine deformation. It is advantageous to get precise estimation. But the feature of local motion is disadvantageous for global motion estimation. In order to improve the accuracy of global motion estimation, an adaptive noise reduction algorithm is presented to eliminate local motion. The parameters of the camera affine model are computed by the least square method. The proposed algorithm is tested by the standard image sequences and compared with other related methods. The experiments show that the proposed algorithm is adaptive and more accurate.