一种林区无人机和地基点云的自动化配准方法

A Method for Automatic Registration of Unmanned Aerial Vehicle and Terrestrial Laser Scanning Point Clouds in Forest Areas

传统的森林资源调查依靠人工野外测量,周期长、劳动强度大;而激光雷达技术的发展为森林资源调查提供了快速、高效的观测手段。地基激光雷达能获取森林样地的详细三维点云数据,但对树冠上层的采样能力不足;无人机的观测视角可获取完整的森林冠层信息,但对林下植被的描述能力有限;二者提供了互补的视角,因此通过对无人机和地基点云进行配准可获取完整的森林三维结构信息,更好地服务于森林资源调查和生态研究。提出了一种林区无人机和地基点云的自动化配准方法,首先分别从无人机和地基点云中提取树的位置;然后以树的位置为配准基元,构建特征三角形,并根据相似性度量,寻找同名三角形对;最后利用同名的三角形对计算无人机和地基点云之间的配准转换参数。实验结果表明,该方法可有效实现林区机载与地面站点云之间的配准,实现了两种观测视角的数据互补,获取了完整的森林三维结构信息,为林业调查和管理提供有力的支撑。

Traditional forest inventories rely on field measurements, which are labor-intensive and time-consuming. The development of Light Detection and Ranging(LiDAR) provides a fast and effective way for forest inventory, and it has become increasingly important in forestry. Terrestrial Laser Scanning(TLS) can collect detailed 3D point clouds of forest plots. However, its ability to collect canopy top is limited. Unmanned Aerial Vehicle(UAV) laser scanning can obtain the integral canopy structure from top view, but its ability to map the understory layer is attenuated. UAV and TLS provides complementary views. The registration of UAV and TLS point clouds can obtain integral forest structure, which facilitates the forest inventory and ecological research greatly. This paper proposed a method for automatic registration of UAV and TLS point clouds.Firstly, tree locations are detected from the UAV and TLS point clouds respectively. Secondly, the tree locations are taken as primitives for registration. Triangles are constructed by the tree locations and their similarities are evaluated for searching the corresponding ones. Thirdly, corresponding triangles are used for the final transformation matrix calculation. Experiment results show that the proposed method can register UAV and TLS point clouds effectively. The fused point clouds are complementary and integral, which can offer support for forest inventories and managements.