基于机器视觉的曲面铸件的缺陷检测技术研究  

Research on Surface of Casting Defect Detection Based on Machine Vision Technology

作　　者：徐建亮[1] 毛建辉[1] 周明安[1] 祝惠一[1] 

机构地区：[1]衢州职业技术学院,浙江 衢州

出　　处：《计算机科学与应用》2020年第1期44-50,共7页Computer Science and Application

基　　金：2018年度浙江省教育厅高校访问工程师“校企合作项目”(No: FG2018160)、衢州科技计划项目(No: 2019k19, 2019006)和“2017年浙江省高职高专院校专业带头人培养对象人才项目”.

摘　　要：传统的工业生产制造,由于科学技术的限制仍然主要采用人工检测的方法去检测产品表面的缺陷,五金产业在抛光研磨制程后的表面研磨管均是由人工进行目检,主要原因为目前的大范围曲面轮廓检测还是以接触式为主,且国际大厂的曲面检或量测解决方案报价多较为昂贵。对铸件研磨过程中产生的动信号进行分析研究,系统把采集到的复杂振动信号转化成简单电信号,对信号进行EMD分析处理。随着研磨时间加长,研磨振动幅度越来越小。在相同实验条件下,该方法较傅里叶波形分析方法更加直观方便。In traditional industrial manufacturing, defects on the product surface are detected using artificial detection methods due to technical constraints. The surface inspection of water hardware after the grinding process is carried out visually and manually because contact method is still the main method to detect surface contour and automatic inspection solutions available in the market for large-scale curved contour inspection are very costly. The complex vibration signal generated in the process of casting grinding was collected and transformed into electronica signal that was analyzed by EMD. With the increasing lapping time, the vibration amplitudes will become smaller. EMD is more effective and intuition than Fast Fourier Transform Algorithm with the same experiment conditions.

关 键 词：缺陷检测 立体视觉 曲面轮廓 机器视觉 

分 类 号：TP3[自动化与计算机技术—计算机科学与技术]