摘要
利用数字图像处理技术,设计出一种基于线阵CCD成像技术的轴承表面缺陷非接触在线检测系统。首先讨论了系统中灯光、摄像机的安装位置对成像效果的影响,然后提出了一种有效的利用麻点呈黑色针孔状凹坑、有一定深度、个别存在或密集分布的图像特征,对轴承外圈麻点缺陷进行的检测方法。该方法可以克服油渍在图像处理中会造成较大误差的缺点,并且能够对麻点缺陷区域进行确定。现场测试表明,该系统抗干扰性较强,精确度较高,而且应用浮动阈值检测法可以使得该算法有较强的鲁棒性,具有一定的实用价值。
Based on CCD image measuring and image processing techniques, an online non-touch measuring system for measuring bearing pock is developed. The effects of system lighting and installing location of cameras on the imaging effect are emphatically discussed, and a novel pock measuring algorithm is brought forward based on image feature that pock is black and pinpointlike pits with a certain depth, present individually or in a cluster. The method can overcome the defect that the oil stain could make a big error in the image processing, and can mark the region of the defect. Experiments indicate that the system has a strong ability of anti-interference, high inspecting precision and strong robustness for float threshold method. The study has a certain practical value.
出处
《铸造技术》
EI
CAS
北大核心
2006年第9期978-980,共3页
Foundry Technology
关键词
图像分割
表面缺陷
货车滚动轴承
机器视觉
Image segmentation
Surface defect
Freight rolling bearing
Computer vision
