中文摘要：

图像匹配是PCB板缺陷检测中的一个重要环节,匹配效果的好坏直接影响着系统的检测结果和检测精度;为了取得良好的匹配效果和降低对系统硬件的精度要求,文章采用了SIFT匹配算法来进行图像匹配;首先应用SIFT算法获取匹配结果,然后计算出精确变换矩阵进行像素异或运算,最后运用数学形态学的方法消除边缘噪声;实验结果表明SIFT算法能够对存在畸变的图像取得良好的匹配效果,准确检测出缺陷。

英文摘要：

Image matching is an important step in detection of PCB defects, the matching result affect the system＇s test results and test accuracy directly. To achieve good matching results and reduce the accuracy requirements of the system hardware, used SIFT matching al- gorithm to do image matching. First applied SIFT algorithm to get matching result, then calculated the exact transformation matrix to do XOR operation of pixels, and finally used mathematical morphological to eliminate marginal noise. Experiment shows that SIFT algorithm can obtain a good matching result for the distorted image, and detect defect accurately.