基于双三次插值算法的集成电路在线检测快速模板匹配

为了解决目前电路板缺陷检测视觉识别系统对多目标识别检测过程实时性差、识别效率低,甚至出现无法识别等问题,首先对集成电路板图像按一定比例进行双三次插值算法处理,并在该基础上创建模板图像,然后进行归一化积相关匹配快速地识别检测出目标的大概位置,最后以双三次插值算法缩放比例值还原目标在原图像中的位置,从而精确识别检测出的目标。实验结果表明:该算法能达到100%识别,运算时间比传统算法快2～11倍,可见该技术提高了检测效率,具有很好的实用性。

Fast Template Matching for Integrated Circuit Online Detection Based on Bicubic Interpolation Algorithm

This project is aimed at solving problems in the current circuit board defect detection visual recognition system, such as poor real-time performance in multi-target recognition detection,low recognition efficiency, and even cases of being unrecognizable. Firstly, the algorithm performed bicubic interpolation algorithm on the integrated circuit board image by a certain ratio, and created a template image on the basis of this, and then performed normalized product correlation matching to quickly identify the approximate position of the detected target. Finally, the bicubic interpolation algorithm is used to scale the value to restore the position of the target in the original image, thereby accurately identifying the detected target. The experimental results show that the algorithm can achieve 100% recognition and the operation time is 11 times faster than the traditional algorithm. It can be seen that the technology improves the detection efficiency and has good practicability.