Estimation Of Canopy Density And Tree Height Based On Uav

The rise of uav technology and its wide application in different fields make it possible to investigate factors related to forestry resources through uav.Based on the method the drones investigated this paper raised,through the digital cameras and sensors on the drones,obtain the images and flight parameter,and sent the above data to the server real time through the 4G network.Through the terminal can obtain the forest area real-time image,longitude and latitude.Forestry personnel can directly identify tree species,growth,local growth environment and other conditions from real-time images.The canopy is segmented from the orthophoto image above the forest area,and the canopy density is measured by combining the transect method.The segmentation algorithm in this paper is compared with the common segmentation algorithm and manual extraction,and the precision detection rate and total detection rate of the two algorithms are calculated.Experimental results show that the segmentation method has high accuracy and the error is about 5%.The error rate is calculated by the ratio of the measured canopy closure to the upshot method,and the results show that the canopy closure below 0.9 has a higher precision.In view of the harsh forest environment,when forestry personnel cannot directly enter the forest area to use instruments to measure,this paper implements the method of generating three-dimensional point cloud of forest area with Pix4d software to measure the tree height.The LiDAR software was used to import the generated point cloud data,and the height attribute of the point cloud data was used to divide the ground point cloud from the tree canopy point cloud through the algorithm tool provided in the software.Then the height of the tree was equal to the difference between the mean height of the N point clouds in the tree canopy and the mean height of the M point clouds on the ground.The real tree height measured by the total station and the measured value of this method are calculated,The results show that the relative error mean value is 5.65%,the maximum error is 10.06%,and the minimum error is 4.83%.The data and the actual height values do once,twice,three times polynomial fitting according to correlation,it shows the coefficient R~2 is over 0.96 and the increased number of polynomial coefficients changed little,shows that measured with 3d point cloud of tree height and the actual tree height has a significantly positive correlation.The measurement method proposed by this method has strong maneuverability and ease of use,which can improve the efficiency of forest resources investigation,and at the same time improve the instantaneity of the data,can improve the economic and ecological benefits of forest resources.