利用目前流行的高分辨率可见光无人机遥感影像生成树木冠层高度模型,采用分水岭分割算法提取单木树高的研究具有重要理论和实践意义。以位于云南省富民县的天然云南松纯林为研究对象,通过大疆Phantom 4 Pro无人机获取低空可见光遥感影像,利用Pix4D Mapper对无人机影像进行预处理及三维重建,生成三维点云,利用LiDAR360处理三维点云,构建DSM,DEM并生成CHM;采用分水岭分割算法对不同郁闭度条件下获得的CHM进行单木分割及树高提取,对提取结果进行精度评价。结果表明:分水岭分割算法能够准确分割CHM,利用无人机可见光遥感影像进行单木树高提取是可行的;将基于无人机可见光影像提取的树高值与野外实地调查得到的树高值进行对比,R²为0.893,RMSE为1.23m,估测精度为87.58%;同时,林分郁闭度会对单木树高估测产生影响,根据不同郁闭度条件下提取的3组样木树高与实地测量树高的决定系数(R²)分别是0.857,0.939和0.921,RMSE分别为1.450,1.097,0.896m,在低郁闭度林分内树高估测的精度显著高于高郁闭度林分。

The feasibility of using the watershed segmentation algorithm to extract tree height based on the canopy height model(CHM) generated by high-resolution imagery of UAV was discussed.Tree height data were obtained from Pinus yunnanensis Franch in Fuming County of Yunnan Province.Remote sensing imagery in the study area of the near ground was obtained through the DJI Phantom 4 Pro drone remote sensing system.Pix4D Mapper software was used to preprocess UAV images and reconstruct the research area in three dimensions and 3D point cloud.LiDAR360 software was used to process 3D point cloud and build a DSM(Digital Surface Model),DEM(Digital elevation Model)and CHM were generated.Then,for different canopy density of CHM,the watershed segmentation algorithm was used to divide and extract tree height.The accuracies of the results were elevated.Results showed that CHM could be accurately divided by the watershed segmentation algorithm and it was feasible that UAV imagery was used to extract the height of Pinus yunnanensis Franch.The comparative analysis of the data on the height of trees acquired from the UAV image and the field measurement shows that the R ²,RMSE,rRMSE is 0.893,1.23m,12.42%.At the same time,the measurement of tree height was affected by canopy density with the R² and root mean squared error(RMSE)value for the least density being R²=0.939,RMSE=1.097m,for the next one was R²=0.921,RMSE=0.896m,and for the largest was R²=0.857,RMSE=1.450m.The accuracy of the measured height of the larger density was higher than that of the smaller density.