基于点云聚类分析的烟支表面三维缺陷检测  

Three-dimensional defect detection on cigarette surfaces based on point cloud clustering analysis

作　　者：吴庆华[1,2] 沈高建 赵德华 张哲铭 任耀强 WU Qinghua;SHEN Gaojian;ZHAO Dehua;ZHANG Zheming;REN Yaoqiang(School of Mechanical Engineering,Hubei University of Technology,Wuhan430068,China;Key Lab of Modern Manufacture Quality Engineering,Wuhan430068,China;Wuhan Cigarette Factory,China Tobacco Hubei Industrial LLC,Wuhan430040,China)

机构地区：[1]湖北工业大学机械工程学院,武汉430068 [2]现代制造质量工程湖北省重点实验室,武汉430068 [3]湖北中烟工业有限责任公司武汉卷烟厂,武汉430040

出　　处：《包装与食品机械》2024年第4期87-93,共7页Packaging and Food Machinery

基　　金：国家自然科学基金项目(51275158);湖北中烟工业有限责任公司科研项目(2022JSGY4WH2B041)。

摘　　要：针对褶皱、布带纹等烟支表面缺陷在识别中无法量化、主观性强的问题,提出一种基于点云聚类分析的烟支表面缺陷检测方法。采用线结构光扫描采集烟支表面点云数据,利用自适应参数的密度聚类分析方法将缺陷点进行聚类;通过聚类思想将缺陷点云分为不同点簇,计算出聚类后的点簇尺寸,为标准化缺陷的尺寸评估提供可量化的缺陷度量。采集1200支烟支样本的点云数据进行缺陷检测试验,结果表明,所提方法对烟支表面缺陷检测的准确率为98.25%。研究可为烟支表面缺陷检测提供参考。To address the problems of non-quantifiable and subjective identification of surface defects such as wrinkles and crease patterns on cigarettes,a surface defect detection method for cigarettes based on point cloud clustering analysis was proposed.This method employs a line-structured light scanning to collect point cloud data of the cigarette surface,and then the defective points are clustered by using the density clustering analysis method with adaptive parameters;Through the concept of clustering,the defect point clouds are divided into different clusters,and the size of the clusters after clustering is calculated,providing a quantifiable metric for standardized defect size assessment.Point cloud data from 1200 cigarette samples were collected for defect detection experiments.The results show that the accuracy of the proposed method for detecting cigarette surface defects is 98.25%.The research provides a reference for cigarette surface defect detection.

关 键 词：聚类分析 三维缺陷检测 烟支 机器视觉 

分 类 号：TS47[农业科学—烟草工业]