基于机器视觉的透射式薄带针孔检测系统

给出一种基于高亮度单波长LED光源和高速线阵相机的透射式针孔检测系统。光源的透射光穿过带材上的孔洞，在图像上形成孔洞的亮斑区。由于铝板面的强反光性，环境光的干扰以及采集触发信号启动，也会在图像上产生一些亮的异常部分。利用图像分割算法定位出图像异常部分后，采用特征判别分类缺陷，判别其是否为孔洞，最后计算缺陷在带材上的位置。大量现场数据测试表明该实时孔洞检测系统是稳定可靠的。

Pinhole detection system for thin metal strip based on machine vision

This paper proposes a transmissive pinhole detection system, which is configured with the high brightness, single- wavelength LEE） light source and the high-speed line scan CCE） camera. The light from the light source passes through the holes on the strip and form a bright spots in the image acquired by the CCD camera. There are also light pseudo defects in the image due to the strong reflection of the aluminum surface, the ambient light and the start of the trigger signals for image acquisition. The anomalous bright area is detected by an image segmentation algorithm, and then a discrimination method is used to classify the detected defects as real pinhole defect and the false defects. Finally,the pinhole defect position on the strip is computed. Large amounts of site data tests show that the real-time pinhole detection system is stable and reliable.