摘要:
利用TLS测量森林参数(树高、胸径和林分密度等)不仅节省人力，而且还能提高工作效率，目前已成为快速获取树木几何参数的有效方法之一。本文利用地基激光雷达数据，提出了一种半自动方式提取单木参数的方法。首先，利用发射脉冲与接收回波脉冲的形状偏差和强度对原始点云数据中的“飞点”和噪点进行滤除；然后，配准各站点云数据；接下来采用不同目标回波过滤方法滤除部分树叶点和其他非树干点；抽稀、合并各站点云数据，对合并后的点云数据水平分层并生成对应的灰度影像；对灰度影像采用Hough变换方法估测单木位置和胸径；对Hough变换检测层数较少的结果采用椭圆拟合方法重新检测；最后，结合Hough变换和椭圆拟合检测结果从点云数据中提取单木位置、胸径和树高。用小兴安岭凉水自然保护区的白桦天然次生林对算法进行了检验,5块多站扫描样地单木识别的平均精度为72.16%，将所提取的单木位置、胸径和树高与外业实际测量数据对比，结果一致性较高。5块多站扫描样地的胸径和树高均方根误差分别为2.38、2.55、3.58、2.21、1.92 cm和4.31、3.87、2.34、5.00、3.47 m。
Abstract:
Using TLS to measure forest parameters (tree height,DBH and tree density,etc.) can not only save manpower, but also improve work efficiency, which has been a fast and effective way to obtain geometric parameters of trees now. A semiautomatic way was employed to extract the individual tree parameters from TLS observations in this paper. Specifically, the noises and “flyingpoints” in the original point clouds were first filtered out by means of intensity and shape deviations between the received echo pulses and the emitted pulses, followed by registration of all stations data, and elimination of partial foliage points and other nontrunk points via the filtering method of different targets echo. Diluting and merging all stations data, horizontal stratification of the merged point clouds to generate the grayscale images were also involved in the current analysis. Using the Hough transform method to estimate locations and DBHs of individual trees, redetecting those fewer results from the Hough transform by means of the ellipse fitting algorithms were also elaborated. Finally, locations, DBHs and heights of individual trees were derived from combining the detection results of Hough transform and ones of ellipse fitting. The average accuracy of individual trees in five multiscans test plots was identified at 72.16%. Comparing the extracted individual tree locations, DBHs and heights with the ground truths, a high consistency between the two data sets was obtained. The RMSEs of DBHs and tree heights in five multiscans plots were 2.38, 2.55, 3.58, 2.21 and 1.92 cm, and 4.31, 3.87, 2.34, 5.00 and 3.47 m respectively.
百度学术
下载:
