【作者】 张健；

【导师】 郑嫦娥； 田野；

【作者基本信息】 北京林业大学 ， 工程（专业学位）， 2019， 硕士

【摘要】 在林地信息测量中,树木胸径的测量是一个重要环节。树木胸径是评价林地质量、生长情况以及林地经营管理的重要数据标准。在当前的树木胸径测量中,仍然主要通过人工来进行测量,这样的方法不仅测量效率低下,测量精度更难以得到保障。为了降低人工劳动强度与测量成本,引入自动化测量方式势在必行。本文着眼于城市公园等林地,针对其特点设计车载胸径测量系统。在目标林地中,测径系统分为测量系统与移动系统。移动系统需要获取林地环境,识别树木的位置,自主规划安全行驶路线,测量系统则获取目标树木的胸径。本文利用激光雷达作为树木点云获取设备,激光雷达与深度摄像机作为导航设备,实现了在城市公园等林地的树木胸径检测与林地内机器人的自主导航。主要研究内容如下:(1)利用激光雷达获取林地环境信息,获取树木的点云数据与距离信息,通过基于密度的DBSCAN聚类算法对点云数据进行聚类。将聚类出的点云簇进行标记,划分出有效数据簇与无效数据簇,对有效数据簇中的数据应用胸径计算算法求出树木的胸径。针对于DBSCAN聚类算法对于距离阈值、邻域样本阈值敏感的缺点,将全局阈值参数改进为根据点到激光雷达的距离所获得的动态阈值参数,使算法对于不同距离,不同密度的点云数据聚类精度更高。(2)在林地中进行自主路径规划的关键在于定位与导航。移动系统通过激光雷达与深度相机获取的环境距离信息、移动系统里程计数据,使用自适应蒙特卡罗定位方法,得到移动系统在二维地图中的位姿。在完成移动系统的定位后,向移动系统发送目标点,移动系统会自主规划安全路径到达目标点。更多还原

【Abstract】 In woodland information measurement,the measurement of tree DBH is an important link.Tree DBH is an important data standard for evaluating the quality,growth and management of woodland.In the current DBH measurement of trees,manual measurement is still the main method,which is not only inefficient,but also difficult to ensure the measurement accuracy.In order to reduce labor intensity and measurement cost,it is imperative to introduce automatic measurement.This paper focuses on urban parks and other woodlands,and designs the vehicle DBH measurement system according to its characteristics.In the target forest,the diameter measuring system is divided into two parts:measuring system and moving system.The mobile system needs to acquire the woodland environment,identify the position of trees,and independently plan the safe driving route,while the measuring system acquires the DBH of the target trees.In this paper,lidar is used as the point cloud acquisition equipment for trees,and lidar and depth camera are used as the navigation equipment to realize the DBD detection of trees in urban parks and other woodlands and the autonomous navigation of robots in woodlands.The main research contents are as follows:(1)The point cloud data and distance information of trees were obtained by using the lidar to obtain the woodland environment information,and the point cloud data were clustered by the density-based DBSCAN clustering algorithm.The point cloud clusters were marked,and the effective data clusters and invalid data clusters were divided.The DBH algorithm was applied to the data in the effective data clusters to calculate the DBH of the trees.Aiming at the disadvantage that DBSCAN clustering algorithm is sensitive to distance threshold and neighborhood sample threshold,the global threshold parameters are improved to dynamic threshold parameters obtained according to the distance from the point to the lidar,so that the clustering accuracy of the algorithm for point cloud data with different distances and densities is higher.(2)The key of autonomous path planning in forestland is location and navigation.The mobile system obtained the environmental distance information and the mileage data of the mobile system through the lidar and depth camera,and used the adaptive monte carlo positioning method to obtain the position and pose of the mobile system in the two-dimensional map.After the positioning of the mobile system is completed,the target point is sent to the mobile system,and the mobile system will independently plan a safe path to reach the target point.更多还原