摘要
针对现有真空注型设备气泡检测和消除方法存在的问题,结合机器视觉和图像处理等技术,研究气泡实时监控方法。首先采用直接图像(DI)法利用CCD数字相机实现真空注型气泡检测,并通过VC++线程技术实现气泡图像连续采集;然后研究适合真空注型气泡图像处理算法,特别针对高重叠、大变形的气泡特征提取,提出一种气泡边缘像素比例算法,较好地解决了重叠气泡的信息丢失问题;最后给出了该方法的具体实现过程,并通过实验,验证其可行性和有效性。
Adopts machine vision and image processing technologies to realize real-time monitoring of bubbles for resolving defec- tion of traditional bubbles detection and elimination methods. Firstly, the Direct Image （DI）method based on CCD digital cameras is used to detect vacuum casting bubbles in real time, and the continuous collection of bubbles images is achieved through thread technology of VC ＋ ＋. Secondly, the image processing algorithms are determined. Especially,the bubble edge pixel proportion algorithm is put forward to extract feature of deformation and overlapping bubbles. It can better resolve information loss problem of bubbles. Finally, the realization process of the method is given. The feasibility of the method is testified by experiments.
出处
《现代制造工程》
CSCD
北大核心
2012年第10期120-124,共5页
Modern Manufacturing Engineering
基金
中国博士后科学基金项目(2011M500755)
上海市重点实验室自主课题项目(ZZ0803)
上海市优秀青年教师选拔专项基金项目(B.37-0109-08-004)
关键词
真空注型
图像处理
气泡监控方法
重叠气泡
vacuum casting
image processing
monitoring of bubbles
overlapping bubbles
