Research And System Design Of Individual Tree Segmentation Algorithm Based On Airborne Lidar

Forest is a significant part of the biosphere and plays an irreplaceable role in maintaining the balance of the globe ecosystem.Accurate measurement of single treerelated parameters is essential for forest management,thus individual tree segmentation is an indispensable part in forestry research.Airborne lidar not only can directly,quickly and accurately obtain the three-dimensional spatial coordinates of the target object,but also has a strong ability of vegetation penetration to obtain the position structure of vegetation interior,which provides technical support for the extraction of forest structure parameters.Based on airborne lidar,the main research contents and results in this paper are as follows:(1)Research on individual tree segmentation algorithm based on airborne lidar data.In this paper,we propose a new individual tree segmentation algorithm using airborne lidar data based on trend discrimination of tree crown point cloud without any prioriknowledge.Because of almost individual tree segmentation algorithms need prior knowledge or set threshold manually,the algorithm proposed in this paper is more applicable.This algorithm is applied to three research areas of Shangkuli farm,Wudaogou tree farm and Alps.The results showed that the algorithm can segment trees effectively,the overall root mean square of extraction rate is 0.95,and the root mean square of F_Score is 0.73.At the same time,the accuracy of this algorithm is also evaluated from three aspects: the segmentation results of different forest form,different forest types and different forest density.It is found that the algorithm has better results in coniferous forest and single forest type with single layer and stand density less than 300 plants/ha.Finally,by using the open benchmark of Alps research area,we compare the algorithm proposed in this paper with nine other individual tree segmentation algorithms,and find that the accuracy and stability of the algorithm proposed in this paper is the best.(2)The individual tree segmentation system based on Web is constructed.Taking the individual tree segmentation proposed in this paper as the core,according to the practical application requirements of individual tree segmentation algorithm,and following the corresponding design principles,an individual tree segmentation system based on Web is designed and developed,which integrates the functions of the three-dimensional display,clipping,filtering,normalization,individual tree segmentation,accuracy evaluation and others.The experimental data are used to test and show the functions of this system.The results show that the system has good interactivity and can satisfied the practical application requirements of individual tree segmentation.