Estimation Method Of Forest Parameters Based On UAV High Resolution Image

Forest is the regulating center of ecological environment,which has many important ecological functions and is the basis of forest resource monitoring.At present,with the rapid development of UAV technology,it has been widely used in forestry investigation according to its advantages of strong mobility and high spatial resolution.With the rapid development of hardware technology,the space resolution of UAV has been greatly improved,which can provide more precise and precise data.Therefore,this paper uses m600pro UAV In 2019,the high-resolution image data of 10 and 11 forest classes in Xiashu forest farm,Jurong City,Zhenjiang City,as well as the survey data of 3 coniferous forests,3 broad-leaved forests and 2mixed coniferous and broad-leaved forests on the ground were obtained as the main data sources,and the remote sensing estimation methods of forest parameters such as tree number,tree height,DBH,biomass were carried out,so as to provide scientific basis for the application of UAV Remote Sensing in forest resource monitoring.Based on the UAV high-resolution image data,this study uses the UAV automatic splicing and modeling to build high-resolution remote sensing image and dense three-dimensional point cloud.Through the geographic modeling of high-resolution image and point cloud data,the forest parameters are extracted,and on this basis,the accuracy is verified by combining the measured data of typical sample plots on the ground.The main conclusions are as follows:（1）Based on agisaft photoscan,the geographic model is constructed by software automatic splicing and three-dimensional dense point cloud.The data of DOM,DSM and CHM in the study area are obtained,and the forest parameters such as number of trees,height of single tree and canopy density are extracted.Through the accuracy verification,the average number detection rate is 0.8558;the average number accuracy rate is 0.8968;the average value of f parameter is 0.8704.It shows that the accuracy of UAV image extraction is high.The height data of the extraction point after grid assignment is the height of the extracted tree crown,and the average determination coefficient R² of the linear fitting model is 0.8879.Therefore,using CHM to extract the height of a single tree can provide a reference for manual measurement.The canopy area and ground area were obtained by masking CHM of typical plots.The canopy closure parameters of typical plots were determined according to the ratio of two band values to plots.The highest and lowest accuracy of canopy closure parameters are 97.5%and 83.33%respectively,and the average accuracy is 92.93%.Therefore,it can be considered that CHM can be used to extract canopy density instead of manual measurement.（2）According to the tree height and DBH of the field survey,the tree height DBH model（H-DBH）was established for three species of broad-leaved forest,coniferous forest and mixed coniferous and broad-leaved forest,and the optimal fitting equation was determined by comparing six models of index function,linear function,logarithmic function,quadratic polynomial function,cubic polynomial function and power function,as well as the determination coefficient R².Finally,the quadratic multiple of broad-leaved forest was obtained The fitting equation R² is 0.8662,the fitting equation R² is 0.8985,and the fitting equation R² is0.757.The estimated value of DBH is obtained by substituting the overestimated value of tree into H-DBH and it is verified by linear regression with the measured value.The average coefficient of determination R² is 0.7741,so it can be considered that the DBH retrieved from UAV data by tree height can replace the actual measurement.（3）According to DBH-M,linear regression was used to verify the measured and estimated biomass of each typical plot.The average determination coefficient of each plot was 0.7596.According to the statistics of the estimated and measured values of single wood biomass of 8plots,the average estimated aboveground biomass of 8 plots is 165.24kg,and the average extracted total aboveground biomass is 4044.37kg.The average biomass accuracy of 8 plots was 90.31%,and the total biomass accuracy was 84.1%.（4）According to the measured and estimated biomass of single tree in the field survey of typical sample plots,the tree height biomass model（H-M）was established for three types of sample plots,and the optimal fitting equation was determined by comparing five kinds of exponential function,linear function,logarithmic function,quadratic polynomial function,cubic polynomial function,power function,six kinds of function models and determining coefficient R² H-M is M=0.3057H^2.4503,R² is 0.8899;H-M is M=0.2077H^2.8164,R² is 0.8927;H-M is M=0.0565H^3.3479,R² is 0.8053.The height biomass model（H-M）of the three tree species was well fitted.Combined with CHM and H-M,the biomass distribution of each forest type area was mapped and analyzed.The estimated results of biomass distribution were basically consistent with the actual situation.The extraction of forest parameters such as the number of trees,tree height,canopy density,and biomass through drone images has the advantages of convenience,efficiency,and efficiency.The extraction of forest tree parameters based on the drone platform is feasible and effective,and can quickly and accurately obtain forest information,especially for artificial The extraction of tree height,plant number and canopy density of the forest has a high reference significance,which can replace the traditional field survey to a certain extent and provide a scientific reference for the application of UAV technology in the monitoring of forest resources at landscape scale.