摘要
针对轴承表面缺陷的位置、面积、深度等随机变化导致图像采集具有不确定性的特点,设计了一套采用光电耦合器图像识别技术进行缺陷自动检测的系统。该系统光源恒定、处理速度快、图像精度高,能长时间稳定工作,显著提高了图像采集质量。经过深入研究图像分割、图像差影、模板匹配等模式识别理论与技术及微小轴承图像特征发现,要检测轴承的表面缺陷,需将轴承表面分割成两部分进行识别,轴承表面的均匀分布区域直接采用差影法,文字刻印区域采用模板匹配法,结果表明该方法识别正确率达98%。
Focusing on the characteristics that the collected image was uncertain because of the bearing surface defect＇s position and area as well as depth changing randomly,an automatic defect inspection system by the image identification technology was designed. The system had advantages of the invariableness of light source and long stable working time, which improved the image collecting pre cision greatly. By studying the digital image features and the theories and technology of pattern recog nition such as segmentation of image, the subtracting between two images and template matching, the paper came to the conclusions that the bearing surface image should be divided into two parts to inspect the defect efficiently, one with regular gray-value and the other with irregular gray-value be cause of model text. The subtraction could be directly used in the former situation and the template matching could be used in the later. Finally the experiment result shows that accuracy of the system reached as high as 98%.
出处
《中国机械工程》
EI
CAS
CSCD
北大核心
2006年第10期1019-1022,共4页
China Mechanical Engineering
基金
宁波市工业科研攻关项目(2005B100014)
关键词
光电耦合器
微小轴承
表面缺陷
差影
模板匹配
change couple device（CCD）
micro bearing
surface defect
substraction
template matching
