摘要
在三维跟踪及逆向工程三维重建的过程中,需要采集大量物体表面的特征点(点云)数据,为达到更好的重建结果,采用多视角采集点云方式。多片点云数据的配准(Registration)是目前的研究热点之一。对点云快速配准方面传统采用的ICP算法加以改进,提出基于相机标定和双向随机KD树的ICP改进算法,提高了ICP在应对小型点云时的匹配精度。
In the three-dimensionaI tracking and 3D reconstruction process,Iarge amount of feature point data （point cIoud） coI ected from different view-point on an object surface are required.The registration of the point cIouds is a key tech-nique.An improved ICP aIgorithm based on bi-direction random KD tree and camera caIibration is introduced in this paper. The experiment shows that the registration accuracy for a pair of smaI point cIouds couId be improved.
出处
《工业控制计算机》
2015年第7期99-101,共3页
Industrial Control Computer
关键词
点云配准
双向随机KD树
相机标定
registration
bi-direction random KD-tree
camera caIibration
