Diameter at breast height of tree is one of the most important factors of tree-measuring, and its precision directly affects the determination of wood volume. The traditional method had low measuring-efficiency, smaller measuring-range. Using remote sensing inversion to measure the tree diameter at breast height indirectly cannot directly access individual tree point cloud data and the precision was relatively lower. The author presented an individual tree diameter algorithm which provides an kind of automatic, efifcient extraction, by using three-dimensional laser scanning technology to extract the 3D point cloud data of tested trees. First of all, by using of 3D laser scanner to scan 8 poplar samples, we got its three-dimensional point cloud data;and then, the cloud data were partitioned, simpliifed and de-noised, thus obtaining the simpliifed point cloud data; ifnally, we layered the extracted data of diameter at breast height of point cloud data sets. The interception layer thickness was set four grades (0 cm, 0~1cm, 1~2cm, 2~3cm;The point cloud data were closed into a polygon ifrst by using fast convex hull algorithm, and then by using the method with ArcEngine controlling ArcGIS to calculate the polygon length, the circumference of a closed plane was calculated, then, the value of tree diameter at breast height was converted out, ifnally combined with the measured data to the traditional algorithm and the circle iftting algorithm, a comparative test and analysis has been carried out. The results show that the traditional algorithm and circle iftting algorithm and fast convex hull algorithm model determination coefifcients R2 were 0.857, 0.941 and 0.957 respectively, and the ifndings explain that individual tree DBH extracted through fast convex hull algorithm is an efifcient and feasible method.%胸径是树木最重要的测树因子之一，其精度直接影响材积的测定。传统的树木胸径测量效率低，范围较小；采用遥感反演间接测量胸径，精度较低，且不能直接获取单木的点云数据。本文利用三维激光扫描技术提取立木的3D点云数据，提出一种自动、高效提取单木胸径的算法。利用三维激光扫描仪对样地8棵杨树进行扫描，得到三维点云数据；同时，开展数据分割、精简、降噪处理，得到简化后的点云数据，最后对提取的胸径点云数据进行分层设置，将截取层厚度设置为0，0～1，1～2，2～3 cm 4个等级，利用快速凸包算法将点云数据闭合成一个多边形，运用ArcEngine控件调用ArcGIS中测算多边形长度的方法计算闭合平面周长，换算出立木胸径值，并结合同步实测数据与传统算法、拟合圆算法进行对比试验。结果表明：采用传统算法、拟合圆算法和快速凸包算法的模型决定系数R2分别为0.857、0.941和0.957，说明运用快速凸包算法提取立木胸径是一种高效且比较可行的方法。
展开▼
