作　　者： (裴新权）; (周亚丽）; (张奇志）; (刘俊）;

机构地区： 北京信息科技大学自动化学院,北京100192

出　　处： 《实验室研究与探索》 2017年第8期24-28,共5页

摘　　要： 在印刷电路板(PCB)裸板质量检测中,针对单次拍摄完整PCB裸板无法获取高分辨率、高精度图像的问题,根据机器视觉原理和数字图像处理技术设计并实现了一种针对PCB裸板质检的图像拼接算法。该算法首先利用SURF算子实现图像特征提取,然后在随机抽样一致算法的基础上优化匹配结果,最后采用保留PCB缺陷特征的改进加权融合方法进行图像融合,完成视觉系统中的PCB图像拼接。实验结果表明,该算法可以有效地对PCB裸板图像进行无缝拼接,同时能够将未拼接图像因精度低而漏检的瑕疵部分清晰展现并输出检测结果,以备后续对PCB裸板进行综合质检,可应用在公司实际产品检测中,具有良好的工程实用性。 For the quality inspection of the bare printed circuit board（ PCB）,according to the requirements for high resolution and high accuracy of image acquisition,an image stitching algorithm for quality inspection of PCB bare board was designed and implemented. Firstly,the speed-up robust features（ SURF） algorithm was used to extract image features. Secondly,the random sample consensus（ RANSAC） algorithm was used to eliminate false matching points.Thirdly,the algorithm calculated the transformation relationship between two images. Finally,an improved weighted fusion method for image fusion was used to complete the image stitching. The experimental results show that the algorithm can stitch pictures seamless without missing the bad features in a quick and efficient way for the quality inspection of PCB bare board,and has good engineering practicability.

关 键 词： 图像拼接 印刷电路板质检 特征 视觉检测