Research On Individual Tree Crown Segmentation Algorithm For Simulating Physical Pouring Water By LiDAR

Forests are an important part of maintaining the balance in the natural world,forest resource investigation and protection are the key part of the sustainable development of forests.In order to obtain accurate information on the vertical structure of individual tree,airborne LiDAR(light detection and ranging,LiDAR)technology and individual tree crown segmentation algorithms were invented.They have important significance for extracting individual tree information for single tree structure research,understanding tree growth and sustainable forest management.This research creatively proposes an algorithm for simulating physical pouring water to achieve the individual tree crown segmentation.The algorithm uses a variable window local maximum algorithm and innovatively uses the layered region growth algorithm to simulate physical pouring water with constraint conditions.In the first experiment,the detection rate of crown vertices in Nanning GaoFeng forest park,Guangxi,f scores are greater than 0.71.In the detection of tree crown,the rRMSE is less than 20%,which indicates that it is feasible in the individual tree crown segmentation of Subtropical Forestry.This thesis then proposes an energy function constructed by height differences and gradients to segment the canopy boundary to solve the problems in the simulated physical pour water algorithm,thereby optimizing the treetop extraction and segmentation results of adjacent canopy.Using subtropical forests in southern China as the research area,the effects of the traditional watershed algorithm under different densities(low,medium,and high density),different tree species(Chinses fir,eucalyptus,and Pinus sylvestris)and the algorithm of this study on tree top extraction and crown segmentation were tested.In order to check the accuracy of the results,the segmentation results of the LiDAR data and the manually measured data were compared and verified in the experiment.The results show that:in terms of the average detection rate of treetop detection effect of the algorithm of this study(r=0.90,p=0.84,f=0.86)is better than traditional watershed algorithm(r=0.62,p=0.81,f=0.78).In terms of the average precision of canopy boundary detection,this research algorithm detects(R~2=0.80,RMSE=0.22 m,rRMSE=12.03%)is better than the traditional watershed algorithm(R~2=0.68,RMSE=0.28 m,rRMSE=17.45%).In the treetop detection,the accuracy rate of eucalyptus is slightly lower than that of Pinus bungeana Zucc but the difference between the detection rate of Pinus bungeana Zucc and eucalyptus is small.In crown width detection,the accuracy of eucalyptus and Pinus bungeana Zucc rise with the increase of density,and both tree species have good performance in crown width detection.In subtropical forests,this algorithm can better improve the effect of individual tree crown segmentation,which is of great significance for understanding forest management,tree competition and resource monitoring.In future research,we will prefer to apply this algorithm to large areas and wild forests for testing and statistics to increase its universality.