Study On Method Of Removing Pits Of Forest Canopy Height Model And Extracting Tree Height Information From Airborne Lidar Point Cloud Data

With the increasing use of airborne light detection and ranging(LiDAR)technology in the forestry sector,high-density LiDAR point cloud data provides not only forest-scale forest parameters,but also single-scale forests.The canopy height model(CHM)as its data product directly affect the inversion of forest parameters,but the presence of local pits in CHM hinders the extraction of forest parameter information.In order to optimize the CHM,the local pits need to be removed.Therefore,for the purpose of using LiDAR data to extract tree height information,this paper proposes a method of detecting point cloud data based on robust local weighted regression to optimize the forest canopy height model without pits.This article mainly focuses on the following aspects:(1)Optimize the canopy height model(CHM),aiming at removing the ubiquitous local crater data,so as to achieve the purpose of optimization,in order to obtain high-quality forest parameters.This paper proposes to use local robust weighted regression to perform point-smoothing processing on point cloud data,fill pits(invalid values),and then use a certain interpolation method to generate a digital surface model(DSM),and subtract it from DEM.A normalized height cloud was obtained—by forming CHM after removal of pits.(2)Compare this method with Gaussian filtering method,median filter method,and stratified height maximum method.Verify the data of 30 sample plots in the study area and extract information from the tree and higher levels.The advantages of removing the pit method.(3)Based on the CHM optimization algorithm proposed in this paper,different interpolation methods are used to generate the canopy height model.The data of 30 sample plots in the study area are processed and the tree and higher information is extracted for verification.The Kriging interpolation method is compared,and recently Tree-less CHM extraction of tree height information generated by neighbor interpolation,spline interpolation,and inverse distance weight interpolation.The results of experiments on 30 sample plots in the experimental area of the Qilian Lake basin in Qilian Mountains show that the smooth point cloud data based on robust weighted regression can effectively recover the true canopy area on the CHM,and the CHM is optimized to fill the CHM.Most of the pit data,while retaining the top canopy morphology and original data characteristics,is compared with Gaussian filtering,median filtering,and stratified height to remove pits.There are obvious advantages in the extraction of the effect or the manuscript information.In addition,by contrasting the Kriging interpolation method,the nearest neighbor interpolation method,the spline function interpolation method and the inverse distance weight interpolation method,the inverse distance weights(IDW)Interpolated non-pit CHM also has significant advantages.In short,using this method to optimize the canopy height model(CHM)of airborne LiDAR point cloud data,and use inverse distance weight interpolation method to generate CHM,the information of tree height can be extracted with higher quality.Subsequent establishment of the basis for the estimation accuracy of forest parameters for individual trees or stands.Therefore,the research work in this paper has a certain practical value in the investigation of forest resources.