自动光学检测中的PCBMark校准研究.pdfVIP

自动光学检测中的PCB Mark校准研究 自动光学检测中的PCB Mark校准研究 Alignment adjustment for PCB Mark in Automatic Optical Inspection 陈臣  程俊　张丰 摘 要  针对自动光学检测中的PCB Mark的定位校准问题，提出了一种新的亚像素圆心定位算法。针对最常见的圆形PCB Mark，首先利用广度优先搜索连通域检测提取目标的边缘位置信息；然后进行圆形特征识别，消除非圆形的干扰目标；之 后通过三次样条插值的方法，得到亚像素精度的边缘定位；最后，对亚像素精度的边缘点进行最小二乘法圆拟合，从而实 现了PCB Mark中心的精确定位。实验结果表明，相比于传统的模板匹配方法和Hough变换方法，该算法大大提高了PCB Mark的定位精度，同时有效地减少了计算量和计算时间，具有很好的实时性和鲁棒性。 关键词  PCB Mark；圆形特征识别；广度优先搜索连通域检测；三次样条插值；最小二乘法拟合 ABSTRACT This paper proposes a novel sub-pixel locating algorithm improving the performance of time-consuming and low-precision PCB Mark alignment adjustment in Automatic Optical Inspection based on the Template matching or Hough transform. Firstly, breadth first search-based blob inspection is operated to find the edge point at pixel level; And then, circle feature identification is used to eliminate the non-circle objects. After that, the cubic spline interpolation is calculated to extract the sub-pixel location of the edge. Finally, the sub-pixel circle center location is accomplished with the least square fitting, and the circle center position is also the location of the PCB Mark center with sub-pixel precision. Experimental results demonstrate that, comparing with the method of Template matching and Hough transform, the proposed algorithm can not only increase the locating accuracy significantly, but also reduce computational complexity effectively. KEYWORDS PCB Mark; Circle feature identification; Breadth first search-based blob inspection; Cubic spline interpolation; Least square fitting 1 引言 随着半导体工艺技术的不断发展，表面贴装技术 中很难应用。 (Surface Mounting Technology, SMT)的标准不断 为此，本文提出了一种新的算法，针对最常用的 提高，对印刷电路板(Printed Circuit Board, PCB)的 圆形PCB Mark，利用圆形特征的识别消除非圆形对 检测也愈发追求高精度和高速度，传统的人工目检和 象的干扰，再通过基于亚