| The morphological characteristics of fruit tree canopy directly shape their diverse appearances,and indirectly reflect the growth status and yield potential of fruit trees.Fruit trees with different architectures have different canopy structures.The measurement methods of canopy morphological characteristics of fruit trees mainly includes manual measurement,image based measurement and three-dimensional measurement.Compared with the commonly used manual measurement and image based measurement,the three-dimensional measurement not only has higher accuracy,but also has more complete description of the 3-D structure of fruit tree canopy.There is no quantitative method to evaluate 3D leaf distribution of pear tree canopy with different architectures.Aiming at this problem,this thesis put forward a method using ground-based LIDAR for acquisition of canopy leaf point cloud,quantitatively compared the pear canopy leaf distribution of different architectures and varieties,and parameterized evaluated the canopy leaf distribution of different architectures and varieties.The main contents were as follows:(1)Taking Houshui,Cuiguan and Yuluxiang pear tree as materials,28 pear trees with Y architecture,‘A’ architecture and ‘3 + 1’ architecture were scanned by FARO focus S70 scanner.By using SCENE and Cloud Compare software,the three-dimensional point could models of pear trees were reconstructed and processed,the point clouds of pear canopy were extracted,the 3-D phenotype database of pear canopy were established.(2)Based on the 3-D point clouds of pear canopy,the 2-D heat map of leaf point clouds distribution of pear canopy was drawn by using MATLAB.The 3-D coordinates of pear point cloud were projected to 2-D coordinates,and simplified into 2-D point cloud distribution map of front view,side view and top view,which represented the distribution of pear canopy along the west-east,south-north and horizontal distribution.Using the layer classification function and histcounts2 2-D plane counting function in MATLAB,the distribution of pear leaf point cloud was presented in the form of 2-D grid.A total of 84 2-D leaf point clouds distribution maps of three tree architectures and three varieties in three directions were obtained by drawing heat maps from the 2-D grid.The leaf point cloud distribution along west-east,south – north,as well as top view horizontal distribution were displayed.(3)Through mathematical statistics,the distribution parameters of point clouds of pear canopy leaves were calculated and compared quantitatively.Firstly,based on R language,two-way ANOVA was used to explore the influence of pear variety and tree architecture on canopy size.Secondly,with the help of voxelization,setting the voxel size as x=0.129,y=0.134,z=0.117.Then the pear canopy point cloud model was spatially divided.Then,by calculating the point cloud frequency probability,point cloud distribution expectation,point cloud distribution variance,point cloud distribution covariance and correlation coefficient,the leaf point cloud distribution of pear tree in single canopy was quantitatively evaluated.Finally,the distribution of canopy leaf point cloud of different pear varieties and tree architectures in multiple directions was compared statistically.Through the quantitative analysis of the canopy leaf distribution of three kinds of architectures and three varieties of pear trees,it is concluded that the tree architecture has a significant effect on the canopy width and the centroid height;the variety of pear tree has no significant effect on the structure of pear tree canopy.From the results of covariance and correlation coefficient of canopy point cloud distribution,it can be concluded that the change of leaf distribution along west-east and north-south was positively correlated,and the similarity of leaf distribution was small.This study on pear canopy leaf distribution will lay a solid foundation for further study of pear tree phenotyping. |
PDF Full Text Request