Continuous moving measurements of trunk locations and diameters at breast height with 3D mapping in plain forests

  • SEKI Saimon
    Graduate School of Systems and Information Engineering, University of Tsukuba
  • TSUBOUCHI Takashi
    Faculty of Engineering, Information and System, University of Tsukuba
  • SARATA Shigeru
    Faculty of Engineering, Information and System, University of Tsukuba
  • HARA Yoshitaka
    Future Robotics Technology Center, Chiba Institute of Technology

Bibliographic Information

Other Title
  • 平地林での3次元地図生成による立木位置と胸高直径の連続移動計測

Abstract

<p>The purpose of this research is measurement in plain forest using a 3D-LIDAR. Point clouds of standing trees are acquired by a mobile 3D-LIDAR that is brought by a human who keeps walking in the forest. We have developed the mobile 3D-LIDAR that has wide field of view. 3D forest maps can be generated by LOAM. LOAM is a method of 6DOF matching with rotating 2D scans. Diameters at breast height and trunk locations are measured from the 3D forest map. We consider a way to prevent the error of matching by LOAM from affecting measuring precision of diameters at breast height and trunk locations of each standing tree. At first, a central axis of the trunk is obtained by cylinder fitting to the point clouds of the trunk. Next, scanned points on the trunk of the one scan are projected onto a plane that has a normal vector parallel to the central axis. Finally, the diameters and the central points are obtained by circle fitting to the projected points, and averages of diameters and central points at the same trunk are calculated for diameters at breast height and trunk locations. In experiment, we have evaluated precision of measured diameters at breast height and distances between standing trees among three walking paths for the measurement in plain forest. Target accuracy of diameters at breast height is achieved for 12 out of 14 standing trees. Target accuracy of distances between standing trees is achieved for all standing tree pairs. It takes approximately 30 minutes in a range of 30 m × 10 m area for ten times measurements by the proposed method. This time is shorter than the conventional manual method. Proposed method can make forest measurement more efficient.</p>

Journal

References(7)*help

See more

Details 詳細情報について

Report a problem

Back to top