A Robust Feature-Based Camera Motion Estimation Method

The camera motion that is caused by the moving camera is one of two types of video motion. The feature-based techniques are known to perform more robust with large motion. The research of such a robust feature-based camera motion estimation method is necessary and possible. In this paper, a robust feature-based camera motion estimation method is proposed. The algorithm uses the motion vectors obtained from a set of selected points to calculate the parameters of the camera motion model. It comprises three steps: the detection of feature points, the computation of correspondences between two sets of features, and the motion parameter estimation. This whole process ensures that only the motion model that had the largest number of inliers is returned as result. Experimental results show the proposed method can be successfully provided with the camera motion estimation capability.