点、线相似不变性的城区航空影像与机载激光雷达点云自动配准

发布时间：2018-09-04 09:04

【摘要】：提出一种基于点、线相似不变性的城区航空影像与机载LiDAR点云自动配准算法。首先通过SIFT算子提取点特征并进行粗配准,同时分别基于影像和LiDAR点云提取直线特征;然后利用局部区域点特征与线特征的相似不变性,通过匹配点对搜索匹配直线对;最后采用基于扩展共线方程的2D-3D严密配准模型实现航空影像与LiDAR点云的精配准。本方法的特点是:采取了由粗到精的配准策略,通过点、线相似不变性,将基于强度的配准算法和基于线特征的配准算法有机结合,在较高的自动化程度下实现了影像与点云的精确配准。试验证明,与基于点云强度影像的自动配准算法相比,本文的算法在城市地区能够取得较好的配准结果。
[Abstract]:This paper presents an automatic registration algorithm for urban aerial images and airborne LiDAR point clouds based on point and line similarity invariance. Firstly, the point feature is extracted by SIFT operator and rough registration, and line feature is extracted based on image and LiDAR point cloud, then the matching line pair is searched by matching point pair by using the similarity invariance between local region point feature and line feature. Finally, the 2D-3D strict registration model based on extended collinear equation is used to realize the fine registration of aerial image and LiDAR point cloud. The characteristics of this method are as follows: the registration strategy from coarse to fine is adopted. Through the invariance of point and line similarity, the intensity based registration algorithm and the line feature based registration algorithm are combined organically. In a higher degree of automation, the accurate registration of image and point cloud is realized. The experimental results show that compared with the automatic registration algorithm based on point cloud intensity image, the proposed algorithm can obtain better registration results in urban areas.
【作者单位】：武汉大学遥感信息工程学院;武汉大学测绘遥感信息工程国家重点实验室;中国科学院上海技术物理研究所;
【基金】：国家973计划(2009CB72400401C) 科技支撑课题(2012BAH34B02) 武汉大学研究生自主科研项目(2012213020203;2012213020209)
【分类号】：TN958.98

【参考文献】