Research On New Method Of Land Production Capacity Monitoring Based On UAV-SFM Technology

Crop aboveground biomass is not only one of the important indicators reflecting the growth of crops,but also one of the important indicators reflecting the production capacity of the crops.It is an important reference for realizing accurate crop yield monitoring and forecasting.At present,the extraction and modeling of point cloud data from the structure of motion(SFM)based on Unmanned Aircraft Vehicle(UAV)data is the frontier of precision agriculture estimation,and using crop height to retrieve crop biomass can obtain higher prediction accuracy,its measurement error can reach centimeter level,which provides a new idea for crop biomass inversion.It also provides a new method for monitoring land production capacity.This paper took the wheat experimental site in the dry farming area of Mazichuan Village,Lijiabao Town,Dingxi City,Gansu Province as the research object.Frist,using UAV platform(Phantom 4)with standard digital camera and GPS to obtain high resolution RGB images.Second,using structure from motion(SFM)and multi-view stereopsis(MVS)technology to generate dense point clouds.Then,the digital surface model(DSM)and orthophoto map are made by dense point clouds,and RGB index was obtained from orthophoto.Next,extracting the ground point cloud information of bare and generating digital terrain model(DTM),the canopy height model(CHM)was constructed.After that analysing the relationship between point cloud variable data,UAV visible vegetation index and wheat biomass fresh weight and dry weight,establishing biomass regression inversion model by choosing the appropriate independent variable factor.Finally,the biomass of wheat in large area was estimated.Accordingly,this study proposes a new method for monitoring land production capacity using a single UAV survey.The main work and related research results are as follows:(1)UAV-SFM technology enables the monitoring of land production capacity,which is low-loss,low-risk and capable of rapid monitoring.(2)Evaluated the accuracy of wheat height model based on measured data of sample points.The determinant coefficient R~2 of the wheat crop height model was 0.914,and the root mean square error RMSE was 2.94 cm.Therefore,based on UAV-SFM method,the estimation of crop height of spring wheat in the study area can be realized.(3)Established simple linear regression model(SLR)and Stepwise linear Regression model(SWL)base on variable factors which screened by correlation analysis.Then,inversed the fresh weight and dry weight of spring wheat biomass,and performed accuracy test and evaluation.The results showed that the accuracy of the SWL biomass fresh weight inversion model(R~2=0.794,RMSE=100.64g/m~2)is higher than the SLR biomass fresh weight inversion model(R~2=0.622,RMSE=101.76 g/m~2).(4)Estimated the fresh weight of wheat in each plot by SWL biomass fresh weight inversion model,The results showed that R~2 was 0.704,RMSE was 1.98 kg,indicate the result was good.Therefore,UAV-SFM can be used to monitor the growth of wheat.