改进的点云精确匹配技术  

An Improved Point Cloud Precisely Matching Technology

作　　者：梁新合[1] 宋志真[1] 

机构地区：[1]河南科技大学材料科学与工程学院,河南洛阳471004

出　　处：《装备制造技术》2008年第3期41-42,共2页Equipment Manufacturing Technology

摘　　要：针对点云精确匹配,提出一种改进的ICP算法。首先采用标签法寻找对应的标志点对,计算变换矩阵完成点云的粗匹配,其次采用KD-Tree的空间搜索策略,计算点云之间的重叠部分以缩小ICP算法的迭代区域;然后给出对应点对的约束条件,通过删除错误的对应点对以提高有效性。该算法提高了精匹配的效率和匹配精度,并用实例验证了方案。Based on precisely matching of point cloud, this paper proposes an improved ICP algorithm. Firstly, it search corresponding points by using reference points and complete crude matching of point cloud by calculating matrix. Secondly, spatial queries with KD-Tree are used to calculate overlap point cloud to minimize the region of the ICP iterative algorithm, afterwards, by giving the restricted conditions of corresponding points and deleting the wrong pair of corresponding points to enhance the validity .It improve the efficiency and precision of matching,and also it was proved by example.

关 键 词：点云匹配 ICP算法 对应点 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]