African Conference on Precision Agriculture (AfCPA) Presentation

Development of Canopy Mapping System of Asian pears (Pyrus pyrifolia Naka) Using Terrestrial Laser Scanning

In this paper, the canopy mapping system (CMS) of Asian pears for estimating yield during Bud thinning and Pruning operations using point cloud data was proposed. Bud thinning and Pruning in Asian pear (Pyrus pyrifolia Naka) is necessary to ensure quality and yield but is time-consuming and heavily depends on work knowledge. This study described a method of estimating the number of fruits through the length of a branch based on remote sensing. The CMS would be useful to support more efficient fruit cultivation and provide information for production management. The objectives of this research were 1) to establish a method for measuring point cloud data of pear trees, 2) to validate the number of fruits prediction based on point cloud data analysis. Point cloud data of pruning of the orchard were measured by using a Terrestrial Laser Scanner (TLS; Topcon GLS-2000). In order to install TLS at the same measurement-point every time and shorten the 3D scanning time, a new type of fixing device Ground-station was made. When using a tripod, the data measurement time per scan point is about 30 minutes, but the Ground-station could measure in 10 minutes, 3 times faster than the tripod fixation method. A total of 4 measurement-points were specified for 3D scanning of the pear tree. The number of measured point cloud points is 18,682,993. Then, to measure the length of the branches, tree data were extracted from the point cloud data, and the predicted number of fruits was verified based on 8 pears/m of each branch. For example, a tree branch with an area of 24 m² (number of points 31,975) has a total length of 22.9 m, multiplied by 8 to calculate the estimated number of fruits from 3D scanning data. The predicted number of fruits was 183 and the actual number of fruits was 164. That means it has 19 fewer fruits than the theoretical number of fruits. So, the measured number of fruits/m² was average 6.8, and the Predicted number of fruits/m² was average 7.6. The absolute error of the two results was 0.8/m². If this information can be mapped, it may be possible to adjust the number of flower buds to the theoretical number at the time for flower thinning in the next season. From the above results, the proposed CMS proposed a method to efficiently measure parameters necessary for the management and monitoring of orchards.



Tottori University