电子科技2025，Vol.38Issue(9)：71-78,8.DOI:10.16180/j.cnki.issn1007-7820.2025.09.009
果树枝干点云的在线配准与参数估计
Online Registration and Parameter Estimation of Point Clouds from Fruit Tree Branches and Trunks
摘要
Abstract
In the process of mechanical vibration harvesting,it is difficult to obtain the three-dimensional shape and parameters of fruit branches online,so the selection of grabbing location and exciting frequency depends on manual experience.To solve this problem,a method based on portable RGB-D(Red-Green-Blue-Depth)camera is proposed to quickly scan fruit tree branches and realize online point cloud registration and 3D(Three-Dimensional)parameter estimation.Multiple sets of color and depth maps of fruit tree branches are collected by depth camera,from which tree branches were separated and point clouds are generated.The feature points in each frame color graph are extracted,the BA(Bundle Adjustment)problem is established and the graph optimization theory is introduced to solve it,and the fruit tree point cloud is registered.The fruit tree skeleton is obtained by twoL1-median skeleton contrac-tions to the fruit tree point cloud.Combined with the point cloud and skeleton of fruit trees,the radius of branches is estimated by the method of chord length estimation.The experimental results show that the registration error can be less than 5 mm,which is less than 45 mm without optimization.The absolute error of all calculated branch radius is less than 6 mm,and the relative error is less than 10％.The proposed method can realize 3D reconstruction of fruit tree branches online and provide parameter support for adaptive vibration harvesting.关键词
林果振动采收/点云配准/捆集优化/图优化/L1-中值/树骨架/弦长估计法/枝干半径估算Key words
vibration harvesting of forest fruit/point cloud registration/bundle optimization/graph optimization/L1-median/tree skeleton/string length estimation method/estimation of stem radius分类
信息技术与安全科学引用本文复制引用
金存献,贺磊盈..果树枝干点云的在线配准与参数估计[J].电子科技,2025,38(9):71-78,8.基金项目
浙江省"领雁"研发攻关计划(2022C02057)Zhejiang"Lingyan"R&D Project(2022C02057) （2022C02057）
