【作者】 宋昀岑； 叶玉堂； 罗颖； 陈镇龙；

【Author】 Yuncen Song Yutang Ye Ying Luo Zhenlong Chen(School of Opto-electronic Information,University of Electronic Science and Technology of China,610054)

【机构】 电子科技大学光电信息学院；

【摘要】 设计了一种用于印制电路板(PCB)光学自动检测系统中的图像配准算法。该算法利用采集到的 PCB 彩色图像,计算其彩色梯度值,从而得到图像中的边缘,再利用 Hausdorff 距离测度将边缘图像与标准图像(reference image) 比较,得到待测图与标准图坐标的一一对应关系,进而确定仿射变换参数,实现图像的配准。该方法避免了极为耗时的自相关计算,并且有着十分理想的定位精度,是一种切实可行的工程化方法。更多还原

【Abstract】 A registration algorithm used in automatic optical inspection(AOI)of printed circuit board(PCB)is proposed. In the first step of this algorithm,the gradient magnitude and its direction of the PCB’s color image are detected,which will be used to get the edges of the original image.Then these edges are compared with the reference image using Hausdorff distance as the similarity measure.After that the feature correspondence between original image and reference image is established.At last,the parameters of the misalignment can be estimated according to these corresponding edge features.Comparing to the correlation-based methods,this algorithm is an efficient algorithm,at the same time it has a quite good precision.So,it is a realistic technique in engineering.更多还原