摘要
针对苹果品质在线分级对外形因素评价的应用需求,提出一种多视角线激光扫描三维视觉测量系统的解决方案。通过空间错位环形分布的6个视角的3组线激光扫描测量模块,采集并拼接融合苹果三维点云数据,计算苹果的横径、纵径、体积等几何量,实现苹果果形分类。实验结果表明,三维点云重建测量苹果横径、纵径、体积的平均测量误差分别为0.157 mm,0.410 mm,2.386 mm 3,采用基于Fréchet距离的苹果果形分类模型得到的果形分类结果与人工目测分类结果一致,实现了外形几何参数和形态的准确测量和分类。
In response to the application requirements of apple s online quality grading for the evaluation of appearance factors,a solution of a multi-view line laser scanning three-dimensional vision measurement system is proposed.Three groups of linear laser scanning measurement modules with six viewing angles distributed in a spatially dislocated ring shape are used to collect and merge three-dimensional point cloud data of apples,calculate the geometrical quantities such as the transverse diameter,longitudinal diameter and volume of the apple,and realize the classification of apple shapes.The results show that the average measurement errors of the reconstructed three-dimensional point cloud to measure the transverse diameter,longitudinal diameter and volume of the apple are 0.157 mm,0.410 mm,and 2.386 mm 3,respectively.The shape classification results obtained by the apple shape classification model based on Fréchet distance are consistent with the manual visual classification results,and the accurate measurement and classification of the geometric parameters and morphology of the shapes are realized.
作者
陈庆光
刘强
张竞成
吴开华
CHEN Qingguang;LIU Qiang;ZHANG Jingcheng;WU Kaihua(School of Automation,Hangzhou Dianzi University,Hangzhou Zhejiang 310018,China)
出处
《杭州电子科技大学学报（自然科学版）》
2022年第4期34-41,56,共9页
Journal of Hangzhou Dianzi University：Natural Sciences
基金
浙江省领雁研发攻关计划资助项目(2022C02026)
国家重点研发计划国际合作资助项目(2019YFE0125300)。
关键词
线激光扫描
三维成像
视觉测量
相机标定
linear laser scanning
three-dimensional imaging
visual measurement
camera calibration
