Three-dimensional Segmentation Of Single Wood In Power Line Corridor Based On Airborne LiDAR Data

In the research of forest resources investigation,urban planning and construction,power network setting and basic road surveying,etc.,it is necessary to take the accurate acquisition of 3D surface terrain information as a prerequisite.The traditional ground measurement methods and photogrammetry methods are greatly affected by clouds and rain,and are easily limited by signal saturation,which brings difficulties to the extraction of various forest parameters.As an emerging active remote sensing technology,Lidar(Light Detection and Ranging,Li DAR)can not only overcome the above limitations,but also has the advantages of short data production cycle and high accuracy,so it has achieved rapid development in recent years.However,at present,the 3D information extraction of forest single trees based on airborne Li DAR point cloud data still has the problem of low accuracy of extracting the information of single tree structure in areas with dense forest trees and canopy.In response to this problem,this paper proposes to use the Canopy Height Model(CHM)to split the single tree information as a priori knowledge,and then use the normalized cut(Normalized cut,Ncut)method to further detect the missing trees.The method greatly improves the accuracy of single tree segmentation,and supports forest resources survey and urban planning and construction.The specific research content includes the following aspects:(1)In view of the inadequacy of the existing filtering method in complex terrain,the concept of isoline is introduced,the traditional progressive encryption triangular mesh filtering method is improved,and the point cloud data is segmented according to the generated isoline information.Select the corresponding parameters of the point cloud data in each area to perform progressive encryption triangle filter processing,which improves the accuracy of point cloud filtering.(2)Different types of point cloud data are classified by different methods,and finally the target category is divided into vegetation points,transmission lines and transmission tower points and building points,and the accurate classification of point clouds is quickly and efficiently realized as a follow-up The basis of single wood extraction.(3)Interpolate the classified vegetation point cloud data to generate a Digital Elevation Model(DEM)and a Digital Surface Model(DSM),generate CHM based on the difference between the two,and use DEM data Elevation normalization was carried out to eliminate the impact of the topographic relief of the study area on the subsequent extraction of individual logs.(4)The single-wood canopy segmentation is performed on the CHM using the local maximum algorithm,and the segmentation results are used as the basis for the subsequent single-wood extraction,and the single-wood canopy detection is performed on the Li DAR point cloud data using the Ncut method.The effect of extracting single trees using only CHM and combining CHM and Ncut methods is compared and analyzed.Accuracy evaluation shows that the latter can effectively improve the detection accuracy and provide a three-dimensional model of single trees and estimate forest parameters in the future.Foundation.