Fig. 1. Flow chart of proposed method

Fig. 2. Six delete templates in upper direction

Fig. 3. Flow chart of thinning algorithm

Fig. 4. Connection of skeleton nodes

Fig. 5. Generated skeletons of Ginkgo tree and Amygdalus triloba f. multiplex tree. (a) Picture of Ginkgo on left and picture of Amygdalus triloba f. multiplex on right; (b) point cloud data of Ginkgo on left and point cloud data of Amygdalus triloba f. multiplex on right; (c) generated skeleton of Ginkgo on left and generated skeleton of Amygdalus triloba f. multiplex on right

Fig. 6. Skeletons of Ginkgo tree generated by point clouds at various angular resolutions. (a) 0.02°; (b) 0.05°; (c) 0.10°

Fig. 7. Generated skeletons of Amygdalus triloba f. multiplex tree with different partition specifications. (a) (40, 40, 40); (b) (60, 60, 60); (c) (80, 80, 80); (d) (100, 100, 100)

Fig. 8. Generated skeletons of Amygdalus triloba f. multiplex tree with different filtering parameters. (a) (0,4); (b) (0,2); (c) (10,0); (d) (20,0)

Fig. 9. Tree skeletons of Ginkgo tree and Amygdalus triloba f. multiplex tree generated by GSA and IVTA methods. (a) Skeleton of Ginkgo tree generated by GSA method; (b) skeleton of Amygdalus triloba f. multiplex tree generated by GSA method; (c) skeleton of Ginkgo tree generated by IVTA method; (d) skeleton of Amygdalus triloba f. multiplex tree generated by IVTA method

Table 1. Acquisition information of point cloud data

Table 2. Skeleton information of Ginkgo tree with different angular resolutions

Table 3. Skeleton information of Amygdalus triloba f. multiplex tree with different partition specifications

Table 4. Skeleton information of Amygdalus triloba f. multiplex tree with different filtering parameters

Table 5. Tree skeleton informations generated by GSA and IVTA methods