| Forest resource information generally refers to the stand factor,including forest type,volume,tree height,diameter and canopy density.The traditional approach to obtaining the forest factor is the field survey,which often takes a long time,a large amount of human and financial support,in addition the mapping cycle is proportional to the surveying area.At present,the traditional surveying and mapping method is being replaced by the new surveying technology.Close-range photogrammetry and airborne LiDAR technology are non-contact three-dimensional coordinate acquisition technology,which can obtain the three-dimensional coordinate information of the target quickly and accurately.Combining with the theory of computer vision,two techniques have been widely used in forestry research,such as carbon storage,biomass,leaf area index,wind evaluation and so on.In this paper,the artificial mixed forest and the complex terrain pure forest as the research object,research based on the close-range photogrammetry technology and airborne lidar technology of the forest parameter inversion work,the main contents are divided into the following parts:Based on the close-range photogrammetry technique,proposing a point cloud extraction algorithm for single tree based on spatial structure.The x and y coordinates of the point cloud data obtained from the close-range photogrammetry technology are converted to the row and column coordinates of the gray-scale image,and the height value is converted to the intensity of the image pixel,thus the forest points cloud data are converted into images;Then the vertex position of each tree is obtained according to the local Maxima principle,and the tree height is obtained according to the height difference between the vertex and the ground.Then the edge information of the canopy is extracted according to the height variation trend in the eight directions of the crown,as a result the point cloud of single tree is extracted,according to the ruled and column width of the tree point cloud,the crown amplitude can be obtained,and the diameter at breast height is calculated according to the empirical equation,finally the accuracy of the experiment result is verified by the data of the external industry measurement.The experimental results show that the R-square of the tree height is 0.75 and the R-square of the crown breadth is 0.61.Based on the airborne LiDAR technique,proposing a point cloud extraction algorithm based on gradient variation for single tree.In the same way,the x and y coordinates of the point cloud data obtained by airborne LiDAR are meshing firstly,acquiring the highest and lowest points in each grid,respectively,then converting to two-dimensional grayscale images I_h and I_l,I_h is used to extract the point cloud data from a single tree,I_l is used to extract DTM(Digital Terrain Model)data from the experimental sample.Then,the vertex of single tree and the gradient direction matrixes are obtained by the local maximum principle.the region growth to extract the edge information is carried out from each crown vertex,then the point cloud data of the single tree is obtained.Then according to the bilateral filter to smooth denoising,so as to obtain the ground surface information,calculate the vertex and ground height difference which is the tree height;According to the row width and the column width of the tree point cloud,the crown amplitude can be obtained.Finally,we use the data from the external industry to verify the accuracy of the experimental results.The experimental results show that the R-square of the tree height is 0.77 and the R-square of the crown breadth is 0.65.From the point of practical application,this paper studies the accuracy of forest parameter in version in the new forestry surveying and mapping method,aiming to help the forestry surveying and mapping work have more perfect development foreground. |
PDF Full Text Request