基于TLS和ALS融合数据的杨树林木表型参数提取研究

Extraction of Poplar Tree Phenotypic Parameters Based on TLS and ALS Fusion Data

基于激光雷达手段获取杨树林木表型参数的方法,探究树木点云提取更优林木表型参数的能力,为林木经营方案的编制提供有力的数据支撑与参考依据。根据地基激光雷达(TLS)与机载激光雷达(ALS)的融合点云数据,利用几何特征树木骨架、提取不完全模拟水分和养分传输的算法(ISTTWN),建立三维单木模型,并获取杨树单木的胸径、树高以及枝条属性因子。结果表明,融合数据的胸径、树高提取值的平均值均高于TLS数据的提取值,同时提取精度也更高。在所提取的枝条属性因子中,提取精度依次为着枝高度>枝长>弦长>着枝角度>分枝角度>弓高,且融合数据的提取精度更高。从RMSE、MAE、MAPE、R^(2)4个方面对枝条属性因子进行提取精度评价,枝长的拟合度最高,地基和融合点云提取值分别达到0.985、0.989;角度的拟合度相对较低,TLS着枝角度提取值的R^(2)仅0.775,但融合后的着枝角度的拟合度提升明显,达到0.887。基于不同相对着枝高度对提取精度分析,未融合前相对着枝高度0.4~0.6的枝条属性精度最高,而后其精度随着冠层高度的增加而降低,由于融合后的冠层点云密度提高,枝条属性因子的提取精度较融合前显著提升,且在相对着枝高度0.8~1.0时提高最为明显,相对提取精度最低的弓高提高了10.04%。研究认为TLS与ALS融合点云数据后,由于数据之间的相互弥补,有效提高点云密度,在三维树木模型研建中能够显著提高林木表型参数的数据提取精度,其中冠层提取精度提升最为明显。

The objectives of this study were to investigate the method of obtaining tree phenotypic parameters based on Lidar,explore the ability of tree point cloud,extract better tree phenotypic parameters,and provide more powerful data support and reference for the preparation of tree management plan.Based on the fused point cloud data of terrestrial LiDAR scanner(TLS)and airborne LiDAR scanner(ALS),the incomplete simulated water and nutrient transport algorithm(ISTTWN)was used to extract the tree branch skeleton with geometric characteristics.Three-dimensional single tree modeling was established and the DBH,tree height and branch information of individual poplar trees were obtained.The results indicated that due to the combination of ALS point cloud data,the average extraction values of DBH and tree height of the fusion data were higher than those of TLS data,and the accuracy was also higher.Among the branch attribute factors extracted,the extraction accuracy was in the order of branch height>branch length>branch chord length>branch angle>axil angl>branch arch height,and the extraction accuracy of fused data was higher.The accuracy of branch attribute factors was evaluated from RMSE,MAE,MAPE and R^(2).The best fit was obtained for branch length,and the extraction values of ground and fusion point cloud reached 0.985 and 0.989,respectively.The fit degree of angles was relatively low,the branch angle of the foundation data extraction value was only 0.775,but the fit degree of the fused branch angle was significantly improved,reaching 0.887.Based on the extraction accuracy analysis of different relative branch heights,the accuracy of branches with relative branch heights of 0.4-0.6 before fusion was the highest,and then its accuracy decreased with the increase of canopy height.Due to the increase of canopy point cloud density after fusion,the extraction accuracy of branch attribute factors was significantly improved compared with that before fusion.When the relative branch height was 0.8-1.0,the accuracy of branch attribute extraction was the most obvious,and the relative bow height of the lowest extraction accuracy increased by 10.04%.It is concluded that the fusion of TLS and ALS point cloud data can effectively improve the high point cloud density due to the mutual complement between the data,which can significantly improve the data extraction accuracy of tree phenotypic parameters in the development of three-dimensional tree model,especially the canopy extraction accuracy.