Study Of Individual Tree Height Extraction Based On LiDAR And Aero Photograph

The forest which feeds up the mankind is one of the most important resources in the earth.As the consideration for nature protection is enhanced, more and more attention has been paidon the vital significance of forest. The forest parameters which are the direct indexesestimating the quality of forest is of great importance. Since the tree height plays a significantrole in all the parameters, measuring tree height accurately has a substantially practicalmeaning in a large scale. The LiDAR is a modern technique for ground observation, andpredominates in a class by itself for height exploration. Given its strong capacity of exactlyobtaining the woodland DEM data and forest height information, the LiDAR has promoted theapplication of remote sensing technique forward in the forestry. At present, a lot of research hasfocused on the extraction of individual tree height from the LiDAR data. However, most oftheses research only use LiDAR data, the tree height estimated in this way is influencedprofoundly by the forest stand density because of lacking of information of depicting theindividual tree crown edge by LiDAR data itself, and will affect the extraction of the treeheight information.Taking Tie Shan Ping forest farm of Chongqing as test site, using high density small sizefootprint LiDAR data and high resolution digital camera images as data sources, this articleextracted individual canopy shapes and tree heights. The main details are as following: First,ground point was separated from the LiDAR data using Tin filtering, Edge Match filtering,Planar Fit filtering, Z Clip filtering and manual editor; Second, high precision DEM wasproduced by triangle nonlinear interpolation to the ground points data, the altitude from DEMwere checked by field measurement, so we can get accurate heights of the objects bysubtracting DEM from DSM; Third, the orthophoto digital camera image was segmented bymulti-scale segmentation method which based on the color and shape of image, then individualtree canopy shapes were extracted efficiently; Finally, the altitude maximum in every treecanopy shape was calculated by GIS analysis function, then the individual tree height wasestimated from regression equations, which were established from the maximum and measuredtree height. Based on field investigation data, regression model was built from 60 trees, treeheight was Inversed and accuracy was checked from 24 trees. The accuracies are better than86% and the total average accuracy is 93.78%. The accuracy of deciduous trees is higher thanthat of coniferous trees.The results show that high precision individual tree height can beextracted from small size footprint laser date and high resolution digital camera images.