基于骨架的柔性变形FPC断路检测

发布时间：2018-03-22 02:37

本文选题：骨架　切入点：柔性形变　出处：《仪器仪表学报》2017年04期 　论文类型：期刊论文

【摘要】：为了解决柔性印刷电路(FPC)翘曲变形容易导致图像出现柔性形变现象造成当前的FPC断路检测方法性能降低的问题,在建立柔性形变模型、分析形变机理的基础上,提出了一种基于骨架处理策略的检测方案。首先提取预处理后FPC图像的线路骨架,然后将其输入到建立的动态模板尺度空间(DTSS)模型中,以去除骨架的分支噪声、得到‘纯净’骨架;接着,提取上一步骤处理所得骨架的全部端点,并通过形态学操作和局部模板匹配定位线路异形结构区域,剔除该端点集合中处于异形结构区域内的噪声端点,从而得到断路端点集合,再根据集合中成员与图像位置的对应关系识别断路缺陷。在基于实际环境建立的SUT-F1图库上进行了算法效果验证,并与其他经典方法进行了对比分析。结果表明,本文方法具备良好的抗形变性能,检测正确率(CDR)高达99.30%,误检率(FDR)低至2.25%,表明了方法的有效性;相比于其他方法,CDR至少提高了5.03%,FDR最少降低了8.22%,显示了方法的优势,且具有一定的实际应用价值。
[Abstract]:In order to solve the problem that the warpage of flexible printed circuit (FPC) easily leads to the appearance of flexible deformation in the image, which results in the deterioration of the performance of the current FPC open detection method, the flexible deformation model is established and the deformation mechanism is analyzed. A detection scheme based on skeleton processing strategy is proposed. Firstly, the line skeleton of pre-processed FPC image is extracted, and then input into the dynamic template scale space model to remove the branching noise of skeleton. The 'pure' skeleton is obtained. Then, all the endpoints of the skeleton are extracted from the previous step, and the abnormal structure area of the line is located by morphological operation and local template matching. The set of broken end points is obtained by eliminating the noisy endpoints in the special-structured region of the endpoint set. Then, according to the corresponding relation between the members and the image position in the set, the open circuit defects are identified. The algorithm effect is verified on the SUT-F1 map library based on the actual environment, and compared with other classical methods. The results show that, This method has good anti-deformation performance, the detection accuracy is as high as 99.30 and the false detection rate is as low as 2.25, which shows that the method is effective, compared with other methods, the CDR is at least 5.03% higher than that of other methods and the minimum reduction is 8.22%, which shows the advantages of the method. And has certain practical application value.
【作者单位】：沈阳工业大学视觉检测技术研究所;辽宁省机器视觉重点实验室;
【基金】：国家自然科学基金(61271325)项目资助
【分类号】：TN41;TP391.41

【相似文献】