基于机器视觉的注塑件缺陷检测方法  

Machine Vision Based Defect Detection Method for Injection Molded Parts

作　　者：仓传跃 徐志鹏[1] 周彬 CANG Chuanyue;XU Zhipeng;ZHOU Bin(College of Metrology,Measurement and Instrument,China Jiliang University,Hangzhou,Zhejiang 310018,China)

机构地区：[1]中国计量大学,计量测试与仪器学院,浙江杭州310018

出　　处：《塑料》2025年第1期149-153,188,共6页Plastics

摘　　要：熔接线是注塑制品中较为常见的缺陷,为检测该缺陷,采用机器视觉技术设计了一种利用骨架提取和斜率判断的图像检测方法。为了克服横梁反光及背景噪声和纹理的影响,利用高斯滤波和形态学运算消除图像中的反光干扰,并使用二值化突出横梁边缘信息;然后,结合处理后的图像,对其进行骨架提取与切割,并采用仿射变换得到旋正后的各段待测区域;最后,利用直线斜率判断的方式将异常区域作为缺陷检测出来。对采集的400张横梁区域图像进行实验测试,结果表明,该方法对于检测熔接线具有高效性,平均检测时长为1 s,且检测识别率达到97.8%,与直接采用直线检测方法相比,漏检率降低了40%,能够满足实际工厂生产的需要。Weld line is a common defect in injection molded products,in order to detect this defect,a machine vision technology was employed and a detection image method was developed using skeleton extraction and slope judgment.Firstly,to overcome the effects of beam glare,background noise,and texture,Gaussian filtering and morphological operations were used to remove glare interference and to highlight beam-edge information through binarization.Secondly,for the processed image,skeleton extraction and segmentation were performed,and the warped and rectified testing regions were obtained through affine transformation.Finally,using line slope judgment,abnormal regions were identified as defects.The collected 400 beam area images were experimentally tested.The results showed that the method had high efficiency in detecting fusion lines,with an average detection time of 1 second and a detection recognition rate of 97.8%.Compared to directly using the line detection method,the miss rate detection was reduced by 40%.It could meet the needs of actual factory production.

关 键 词：熔接线 机器视觉 缺陷检测 骨架提取 斜率判断 

分 类 号：TP391[自动化与计算机技术—计算机应用技术] TQ320.6[自动化与计算机技术—计算机科学与技术]