109. Novel chestnut tree crowns segmentation method by UAV oblique photogrammetry

Detection of individual trees within extensive remotely sensed data is an essential phase to derive accurate knowledge of crop status, enabling Precision Agriculture (PA) approaches. In this work, a novel method for tree crown segmentation within a 3D point cloud is presented, which allows the location of each individual chestnut to be automatically detected, and to assign point of the model to the proper tree canopy. The algorithm is based on a multi-filtering approach and the watershed transform clustering. A chestnut orchard in Dronero (Piedmont, Italy) was chosen as a case study. The results of this work enable the full exploitation of 3D point cloud model of the crop, allowing further specific processing at individual tree level for extraction of crown information.