摘要
在金锡共晶焊接以及X射线检测缺陷技术广泛使用的背景下,针对PCB板缺陷经X射线成像后不易人工检测的问题,基于形态学方法提出了一种自动检测并标记烧结空洞的处理思路。经实验发现Sobel算子无法有效提取烧结空洞,分析得出以下结论:提取烧结空洞需要克服水平、垂直方向上的干扰,即密集分布于PCB板上的导线。由此提出利用形态学方法来抑制导线干扰,并通过实验验证了该方法的有效性。同时,确定了提取并标记烧结空洞的主要流程:滤波预处理、形态学方法提取图像边缘、填充空洞区域、Otsu法分割图像、形态学方法去除毛刺、腐蚀梯度算子提取空洞内边缘以及标记烧结空洞。经实验提取出完整的空洞边界,成功标记出PCB板上的烧结空洞,证明了该处理流程的可行性。最后,分析该处理思路的优势以及缺陷,并根据缺陷对改进方向作了建议。
With widespread using of gold-tin eutectic welding and X-ray defect detection technology,in view of the problem that PCB board defects are not easy to be manually detected after X-ray imaging,we propose a method based on morphological methods to automatically detect and mark sintering cavities.It is found through experiments that the Sobel operator can not effectively extract sintered cavities,and the following conclusions are drawn:the extraction of sintered cavities needs to overcome the horizontal and vertical interference,the wires densely distributed on the PCB.Therefore,we propose the use of morphological methods to suppress wire interference and verify the effectiveness of this method according to experiments.At the same time,the main processes for extracting and marking sintered cavities are determined:filter preprocessing,extracting image edges by morphological method,filling the cavity area,segmenting the image by Otsu method,removing burrs by morphological method,extracting the inner edge of the cavities by corrosion gradient operator and marking sintered cavities.The complete cavity boundary is extracted through experiments,and the sintered cavities on the PCB board are successfully marked,which proves the feasibility of this method.Finally,we analyze the advantages and disadvantages of this method and make suggestions on the improvement direction according to the disadvantages.
作者
程金纬
邱晓晖
陈青青
CHENG Jin-wei;QIU Xiao-hui;CHEN Qing-qing(Nanjing University of Posts and Telecommunications,Nanjing 210003,China;Nanjing No.55 Research Institute of Ministry of Information Industry,Nanjing 211100,China)
出处
《计算机技术与发展》
2021年第4期63-68,共6页
Computer Technology and Development
基金
江苏省自然科学基金(BK2011789)。
关键词
共晶焊接
PCB图像
X光图像
缺陷标记
形态学方法
边缘检测
eutectic welding
PCB image
X-ray inspection
defect marking
morphological techniques
edge detection
