Acta Optica Sinica, Volume. 29, Issue 11, 3129(2009)
An Extended Orthogonal Iteration Algorithm for Pose Estimation from Point and Line Segment Correspondences
The orthogonal iteration algorithm is fast and globally convergent pose estimation based on points.And it is a one of the state-of-art real-time pose estimation algorithms.A new method that we express the uncertainty of sensed line segment using object-space line segment error is introduced.An extended orthogonal iteration algorithm is proposed.And it can use point and line segment feature simultaneously.Firsly we formulate coplanarity equation of point data into the same mathematics format with collinearity equation of point data and translate object-space coplanarity error of line segment into coplanarity error of point.Then the sum of object-space collinearity error of point and coplanarity error of line segment is used as error function.Finally,an iterative algorithm is developed to minimize the error function.The experiment results show that extended orthogonal iteration algorithm is effective and accurate.The algorithm can give more accurate,stable result and increase the noise resistance performance compared with the orthogonal iteration algorithm.
Get Citation
Copy Citation Text
Xu Yunxi, Jiang Yunliang, Chen Fang. An Extended Orthogonal Iteration Algorithm for Pose Estimation from Point and Line Segment Correspondences[J]. Acta Optica Sinica, 2009, 29(11): 3129
Category: Machine Vision
Received: Nov. 13, 2008
Accepted: --
Published Online: Nov. 16, 2009
The Author Email: Yunxi Xu (xuyunxi@tom.com)