基于多站扫描的点云特征参数与材积结构动态分析

【目的】采用地面激光雷达(TLS)进行多站点扫描获取立木的点云信息,提取有关点云分布的特征参数,拓展立木测树因子,建立基于特征参数的材积模型。【方法】以马褂木(Liriodendron chinense)人工林为研究对象,利用点云数据提供的立木上部直径(d)、树高(H)等因子对两期(2014、2017年)林分结构变化进行分析;设计并提取基于TLS点云的特征参数高度累计百分比,同时提取了其他与高度相关的特征参数作为一组变量;将提取的立木胸径(DBH)与特征参数作为另一组变量;最后分析特征参数、胸径与材积的相关性,通过逐步回归法分别建立基于两组变量的材积模型,并分析两期材积的动态变化。【结果】选用特征参数H₂₅与H_{t, var}(点云高度方差)拟合两期材积模型,其决定系数R²分别为0.771 1、0.742 6;利用特征参数H₂₅与胸径拟合,模型预测精度有明显的提升。以上两组材积模型预测各径阶材积变化,其模型值与实测值无显著差异,R²均高于0.9。【结论】研究提取的高度累计百分比与立木测树因子紧密相关,可以很好地反演林木的动态结构。研究建立的材积模型均有较高的精度,可用于林木材积动态变化监测,为地面激光扫描点云参与森林资源动态监测提供理论参考。

Dynamic analysis of point cloud characteristic parameters and volume structure based on multi-station scan

【Objective】 The use of terrestrial laser scanner (TLS) in multistations to obtain the point cloud information of a standing tree, extracting the characteristic parameters of the point cloud distribution that expand the tree measurement factor, and establishing a volume model based on these characteristic parameters. 【Method】Liriodendron chinense plantation was studied, and the structural changes of the two stages (2014, 2017)were analyzed using the upper diameter (d) and tree height (H) of the standing tree provided by the point cloud data. The characteristic parameters of TLS point cloud named high cumulative percentage was designed and extracted, and other height-related feature parameters were extracted as a set of variables; the extracted DBH and feature parameters were considered as another set of variables; finally, the characteristic parameters were analyzed. The correlation between DBH and volume was established by stepwise regression method to build a volume model based on two sets of variables, and analyze the dynamic changes of the two phases. 【Result】The characteristic parametersH₂₅ and H_{t, var} (point cloud height variance) were used to fit the two-stage volume model, and their R² were 0.771 1 and 0.742 6, respectively. The accuracy of the model using the characteristic parameter H₂₅ and the DBH improved. The two sets of volume models mentioned earlier were used to predict the volume change in each step; the model value and the measured value was not significantly different, and R² is higher than 0.9. 【Conclusion】 The high cumulative percentage extracted in this study is closely related to the tree measurement factor, which can invert the dynamic structure of forest trees. The volume model developed by the research achieves high precision and can be used for the monitoring of the dynamic change in forest wood product, which provides a new reference for TLS point cloud to participate in the dynamic monitoring of forest resources.