基于SIFT算法与Harris角点检测的PCB板Mark点匹配研究

Research on Markpoint Match of PCB Based on Algorithmsof SIFT and Harris Corner Detection

针对SIFT算法不能有效提取PCB图像与Mark点模板图像的关键特征匹配点的问题,笔者提出基于Harris角点检测与SIFT算法的模板匹配算法。在Harris角点检测的基础上,依靠图像金字塔、位置坐标关系确定出存在的特征点,除去不稳定特征点并生成稳定的特征点。对PCB原始图像与分割图像分别进行只依赖于SIFT算法以及结合Harris角点检测的Mark点匹配实验,实验结果表明该算法能够实现PCB分割图像与Mark点模板图像的准确匹配,提升了SIFT算法性能。

SIFT algorithm can't effectively extract feature match point of PCB image and Markpoint template image.In this paper,a template matching algorithm based on Harris and SIFT algorithm is proposed.On basis of Harris corner detection,it relies on image pyramid and position coordinate relationship to determine the existing feature points and remove unstable ones to produce stable ones.It is verified by testing PCB image when only SIFT or SIFT-based Harris comer detection is used.The results indicates this method achieves accurate match between segmented PCB image and template image,and improves the performance of SIFT algorithm.