Fig. 1. Schematic of acquisition method for soybean plants

Fig. 2. Flow chart of three dimensional (3D) reconstruction of soybean canopy with color information. (a) Synchronous acquisition system; (b) depth image; (c) color image; (d) raw 3D point cloud; (e) gray histogram of color image; (f) distance image of raw data; (g) distance image of raw data with color information; (h) distance image of canopy after segmentation; (i) distance image of canopy with color information after segmentation

Fig. 3. Visual model of Kinect camera

Fig. 4. Relative position relationship between checkerboard and camera. (a) Corner detection of checkerboard by color camera; (b) corner detection of checkerboard by depth camera

Fig. 5. 3D reconstruction of soybean canopy with color information. (a) Raw 3D point cloud; (b) raw 3D point cloud with color information; (c) 3D point cloud distance image of soybean canopy after segmentation under front view; (d) 3D point cloud distance image of soybean canopy after segmentation under side view; (e) 3D reconstruction of soybean canopy with color information under front view; (f) 3D reconstruction of soybean canopy with color information under top view

Fig. 6. 3D image acquisition of individual soybean canopy

Fig. 7. Calculation method of plant height for grouped soybean plants

Fig. 8. Schematic of spacing between pots

Fig. 9. 3D reconstruction effect of soybean canopy on July 19, 2018. (a) 3D reconstruction of individual soybean plant; (b) 3D reconstruction of grouped soybean plants

Fig. 10. 3D reconstruction of 9-5 soybean canopy in reproductive period from June 19, 2018 (0619) to September 5, 2018 (0905)

Fig. 11. Comparison of manual measurement values and sensor measurement values of plant height for three soybean varieties. (a) Comparison result of Kangxian-9; (b) comparison result of Kangxian-13; (c) comparison result of Fudou-6

Fig. 12. Dynamic change of mean value of plant height for three soybean varieties based on time sequence