摘要
为了实现对工件的自动影像测量,建立了自动影像测量系统。对该系统所采用的图元识别、图元测量、基于自动聚焦原理的高度测量等算法进行研究。根据圆形和矩形图元的面积和真圆度等特征参数介绍了图元识别算法。以直线和圆形图元为例分析了典型图元的测量算法,即在提取图元边缘点的基础上进行图元拟合。在分析比较能量谱等聚焦评价方法的性能的基础上,说明了采用自动聚焦原理进行高度测量的算法。最后,介绍了系统比例尺的设定方法以及光栅尺读数的误差补偿方法。实验结果表明:比例尺的标定精度为0.5μm;图元的测量精度在微米级;高度测量精度为0.035 mm,基本满足自动影像测量的稳定可靠、精度高、抗干扰能力强等要求。
In order to realize automatic image measurement for work pieces, an automatic image measurement system is established and the algorithms used in this system such as geometric figure recognition and measurement, height measurement based on autofocus, etc are investigated. The geometric figure recognition algorithms according to the geometric features such as the circle and circularity of area and rectangle are presented and measuring principles and algorithms of typical plane geometric figures are analyzed by taking the measurement of the line and circle geometric figures as an example i. e. , the figure fitting is adopted after the edge points of geometric figures are extracted. Then, after the performance of several autofocus evaluating methods is compared, the algorithm of height measurement by visual automatic focusing is analyzed. Finally, the setting method for scale and the error compensation algarithm of raster reading are discussed as well. Experimental results indicate that the prec＇ision of the system＇s scale is 0.5 μm and the measuring precision of figures is within micron, the precision of height measurement is 0. 035 mm, which can satisfy the system requirements of non-contact, online, real time, higher precision and rapid speed, as well as strong anti-jamming and stabilization.
出处
《光学精密工程》
EI
CAS
CSCD
北大核心
2007年第2期294-301,共8页
Optics and Precision Engineering
基金
粤港关键领域重点突破项目(No.2004A10403021)
广东省攻关项目(No.2006A10401006)
关键词
计算机视觉
边缘检测
图元
曲线拟合
亚像素
computer vision
edge detection
geometric figure
curve fitting
subpixel
