基于Hough变换和K-means聚类的玻璃瓶分模线缺陷检测  

Defect Detection of Glass Bottle Parting Line Based on Time-frequency Domain Analysis And K-means Clustering

作　　者：柴子凡 赵丽琴[1] 蒋海波 段能全[1] 马清艳[1] 王栋[1] CHAI Zifan;ZHAO Liqin;JIANG Haibo;DUAN Nengquan;MA Qingyan;WANG Dong(School of Mechanical Engineering,North University of China,Taiyuan 030051,China)

机构地区：[1]中北大学机械工程学院,太原030051

出　　处：《机械设计与研究》2023年第3期171-174,共4页Machine Design And Research

基　　金：2021年教育部产学合作协同育人项目(202102326033)。

摘　　要：对于玻璃瓶分模线不易识别,难以诊断质量的问题,提出了一种基于傅里叶变换与K-means聚类相结合的玻璃瓶分模线质量检测的方法。通过合理布置结构来采集玻璃瓶分模线图像,并在算法的前处理过程中,针对分模线背景的复杂性提出了基于时频域变换的算法;在分模线质量检测的过程中,提出了Hough变换和K-means聚类相结合的方法。结果显示:该方法能有效对玻璃瓶分模线的质量进行诊断,诊断正确率能达到96%。文中所用方法不仅可以应用于玻璃瓶的分模线检测,且在其他玻璃制品的缺陷检测中也有良好的应用。To address the problem of difficult identification and quality diagnosis of glass bottle parting lines,a quality detection method of glass bottle parting line based on Fourier transform and K-means clustering is proposed.In the process of pre-processing,a time-frequency transform algorithm is proposed based on the complexity of the background of the glass parting line.A method combining Hough transform and K-means clustering is proposed in the process of quality inspection of the parting lines.The results show that this method can diagnose the quality of glass parting line effectively,and the diagnostic accuracy can reach 96%.The method used in this paper can be applied not only to the parting line detection of glass bottles,but also to the defect detection of other glass products.

关 键 词：傅里叶变换 霍夫变换 K-MEANS 缺陷检测 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]