基于网格分析的激光点单株果树识别  

Single Fruit Tree Recognition Based on Grid Analysis and Laser Point

作　　者：罗秋慧 李航[1] 杨弘凡 吴丽 LUO Qiu-hui;LI Hang;YANG Hong-fan;WU Li(School of Mechatronics,Henan University of Science and Technology,Luoyang 471003,China;School of Mechatronics,Luoyang Polytechnic,Luoyang 471000,China)

机构地区：[1]河南科技大学机电工程学院,河南洛阳471003 [2]洛阳职业技术学院机电工程学院,河南洛阳471000

出　　处：《软件导刊》2020年第4期194-198,共5页Software Guide

基　　金：河南省科技攻关项目(182102110420)。

摘　　要：为快速识别果园环境中果树的分布情况,提高果园机械自动化果树定位精度,在控制成本与缺乏信息的前提下使用少线数激光雷达,提出一种基于地物点云侧投影网格分析方法实现果树识别。该方法通过线雷达获取果树信息点云,然后根据点云角度信息进行侧面二维投影、地物聚类划分网格、高程跨度计算、逐层分析、逐网判断等,完成果树识别。试验证明,该方法在地物信息缺乏的情况下可实现果树准确识别,低成本地快速判断果园中果树分布情况,是一种有效的果树识别方法。In order to quickly identify the distribution of fruit trees in orchard environment and improve the positioning accuracy of orchard machinery automation,a method based on cloud side projection grid analysis of ground objects is proposed to identify fruit trees by using a few-line LiDAR in the absence of geophysical information on the basis of cost control.This method obtains the point cloud of fruit tree information by line radar,and then identify fruit tree by side two-dimensional projection,ground object clustering,elevation span calculation,layer-by-layer analysis and network-by-network judgment according to the angle information of point cloud.Experiments show that this method can accurately identify fruit trees in the absence of geophysical information,quickly judge the distribution of fruit trees in orchards at low cost,therefore it is an effective method for identifying fruit trees.

关 键 词：果树识别 点云数据 乏信息 地物聚类 网格分析 

分 类 号：TP319[自动化与计算机技术—计算机软件与理论]