本文为澳大利亚塔斯马尼亚大学（作者：Luke Oliver Wallace）的博士论文，共158页。

机载激光扫描（ALS）已成为一种重要的工具，为大面积森林的三维结构提供经济有效的表征。由于数据分辨率通常与覆盖面积成反比，从可替代平台进行激光扫描一直是最近的研究课题。本文通过研究无人机（UAV）作为森林调查激光扫描平台（UAVLS）的应用，提出了这一探索课题。介绍了由激光扫描器、基于微机电系统的惯性测量单元（IMU）、双频全球定位系统（GPS）接收机和用于Oktokopter多旋翼平台向下指向摄像机组成的小型激光扫描系统的设计。提出了一种利用视觉辅助GPS-IMU西格玛点卡尔曼平滑器进行激光回波直接地理参考的新算法。通过随机和实际评估，可以达到与现代ALS系统相似的精度，并足以进行森林调查测量（水平34厘米，垂直14厘米RMSE）。选择塔斯马尼亚东南部两个4年生桉树种植园作为主要研究区域，以评估UAVLS系统在绘制和评估关键库存指标变化方面的效用。对不同飞行参数捕获的点云分析表明，飞行高度应限制在地面以上50米以内，扫描角度应限制在±30°以内。在这些限制条件下，设计了一种利用重叠样条的调查方法，以便对地块大小区域（500 m²）的三维结构进行成本效益高且可重复的观测。结果表明，在重复多次飞行中，各点位描述统计的最大偏差小于3%。使用五种不同的自动树检测和描绘方法，研究从高密度UAVLS点云（高达300个点/m²）导出的单个树级度量的准确性和重复性，强调增加的数据分辨率在单个树特征描述中提供了更详细信息。使用CHM和点云的最佳表现方法，有98%的树木被重复且正确地从点云中描绘出来。与目前使用的现代野外技术相比，从所描绘的树段中提取的树高（绝对平均偏差0.35 m）、位置（0.48 m）、树冠面积（3.3 m²）和郁闭度（2.3%）具有更高的重复性和更好的效率。随后对连续造林处理后的变化进行了分析，结果表明，UAVLS能够检测到位于96%到125%之间的实际修剪率。本文证明，UAVLS提供了前所未有的时间和空间分辨率，能够确定高精度的森林调查指标及其随时间的变化。与现场技术相比，UAVLS能更有效、更详细地描述森林的三维结构。

Airborne Laser scanning (ALS) has emergedas an important tool for providing costeffective characterisation of the 3Dstructure of forests over large areas. As data resolution is often inverselyproportional to coverage area, laser scanning from alternative platforms hasbeen a recent subject of investigation. This thesis advances this explorationby investigating the use of Unmanned Aerial Vehicles (UAVs) as a laser scanningplatform (UAVLS) for forest inventory purposes. The design of a small laser scanningsystem consisting of an automotive laser scanner, a Micro-Electro-MechanicalSystems based Inertial Measurement Unit (IMU), a dual frequency GlobalPositioning System (GPS) receiver and a downward pointing video camera for useon-board an Oktokopter multirotor platform is described. A novel algorithm wasdeveloped for the direct georeferencing of laser returns utilising a visionaided GPS-IMU sigma-point Kalman smoother. Evaluating improvements due to theinclusion of vision, both stochastically and in practice, it is demonstratedthat an accuracy similar to modern ALS systems and adequate for forestinventory measurements can be achieved (34 cm horizontal, 14 cm vertical RMSE).Two 4 year old Eucalyptus plantations in south east Tasmania were selected asthe primary study area in order to assess the utility of the UAVLS system tomap and assess change in key inventory metrics. Analysis of the point cloudscaptured with different flying parameters indicated that the flying heightshould be restricted to less than 50 m above ground level and scan anglerestricted to ±30◦ . A survey method within these restraints and utilisingoverlapping transects was designed to provide cost-effective and repeatableobservations of the 3D structure of the plot sized areas (500 m2 ). It wasfound that the maximum deviations of plot level descriptive statistics capturedin repeat multiple flights were less than 3%. Investigating the accuracy andrepeatability of individual tree level metrics derived from the high density UAVLSpoint clouds (up to 300 points/m2 ) using five different automatic treedetection and delineation methods highlighted that increased data resolutionprovided more detail in the characterisation of individual trees. The bestperforming method, which utilised both the CHM and the point cloud, resulted in98% of trees being repeatedly and correctly delineated from the point cloud.Tree height (absolute mean deviation of 0.35 m), location (0.48 m), crown area(3.3 m2 ) and canopy closure (2.3%) extracted from the delineated tree segmentswere observed with higher repeatability and better efficiency than thatcurrently achieved using modern field techniques. Subsequent analysis of changefollowing the application of sequential silvicultural treatments showed thatUAVLS is capable of detecting pruning rates of between 96 and 125% of the truepruning rate. This thesis demonstrates that UAVLS offers unprecedented temporaland spatial resolution, enabling the determination of highly accurate forestinventory metrics and their change over time. In comparison with in situ fieldtechniques, UAVLS offers more efficient and detailed characterisation of the 3Dstructure of forests.

1. 引言

2. 超时空无人机激光扫描森林调查误差评估与抑制

3. 无人机激光扫描系统在森林调查中的应用

4. 超高分辨率无人机激光扫描数据的树木检测与分割程序评估

更多精彩文章请关注公众号：