基于李群代数表达帧间位姿变化矩阵的3D视觉跟踪研究

3D Visual Tracking Based on the SE(3) Group Representation of Interframe Pose Transform Matrix

基于RAPiD边缘跟踪方法的思想,采用以帧间位姿变化更新目标物绝对位姿值的办法,引入李群代数表达帧间位姿变化矩阵;然后将帧间位姿变化矩阵参数的求解问题转换为线性最小二乘的最优求解问题.在李群代数的表达方式下,最小二乘的最优问题可较容易获得线性化表达,方便了求解.实验显示了该算法的3D跟踪效果与目标物位姿值的估计结果.

The SE（3） group is introduced for representing the interframe pose transform which can be used for updating the absolute pose of the object.Then,the 3D tracking problem can be equivalent to the least square optimization problem on parameter solving of the interframe pose transform.Owe to the SE（3） group representation,simpler linear representation can be used in optimization process,which pave the way for parameter solving.Experiments show the 3D tracking results of an object and the related pose estimation.