摘要
开发了一套晶振外壳抽检系统,通过对晶振外壳这种微型冲压件的缺陷检测,实现对冲压模具磨损情况的无人监控。在处理算法上,充分针对目标特点,综合运用了边缘检测、数学形态学、拓扑描述等数字图像处理工具,克服了微小金属元件表面反光不均匀,冲压件纹理无规律的影响。实验显示系统漏检率和误检率小于0.50%,完全可以满足抽检系统对于准确性的要求,且成本低廉,为实用化、商品化奠定了基础。
A machine - vision spot - inspection system for crystal oscillator shell （ COS ） quality detection was developed, which can monitor the punch mould state by spot -inspecting the COS automatically. Aiming at the characteristic of the COS, some algorithms including canny edge detector, mathematical morphology, topological descriptors were used to overcome the disturbances. The experimental results showed that the system was able to inspect the COS at a rate of 4 COSs/m and the inaccurate classification rate was no more than 0.50%. The system has commercial utilization for its low - price and high - accuracy.
出处
《制造技术与机床》
CSCD
北大核心
2007年第3期87-91,共5页
Manufacturing Technology & Machine Tool
关键词
视觉检测系统
磨损检测
数学形态学
拓扑描述
Visuion Inspection System
Abrasion Detection
Morphology
Topological Description
