一种印刷电路板锥束CT图像中过孔定位检测方法

印刷电路板（print circuit board，PCB）广泛应用在各种精密仪器中，PCB的质量直接影响器件的整体性能，因此在工业生产中需要对PCB的质量进行检测。过孔负责PCB中各电路层之间的连通，对PCB意义重大。针对PCB的过孔数量大、目标小和形状不规则等问题，结合三维空间中过孔的形态特征设计了一种基于加权叠加增强的PCB过孔检测算法。该算法首先在垂直PCB方向上进行加权叠加增强过孔目标，降低非过孔要素对过孔检测的影响，然后引入一种高斯加权计算区域圆度的方法，利用圆度判断过孔区域，并利用区域信息计算过孔属性。实验结果表明，相比于传统的Hough变换，本方法检测精度高、稳定性强，同时大大降低了人工检测的工作量。

Detection Via Location in the PCB Cone Beam CT Image

Print circuit boards（PCB）are widely used in all kinds of precision tools. The quality of PCB affects the whole instrument capability directly. Therefore, the detection of PCB quality is needed in the industry production. Via connects the different slices of the PCB, and is very important to the PCB. To solve the problem that the number of vias is huge, the target is small and the morpho- logic is abnormal, an algorithm based on regional roundness combining the morphologic character of via in the 3D space is designed to detect the vias in the CT image of PCB. First, every slice data in the vertical direction of the PCB is superimposed to enhanced vias target, and to reduce non-vias goal. Then a new roundness is defined which is used to recognize the region of vias, meanwhile the property of the region is obtained, from which the property of via can be calculated. Experimental results show that the method is good in veracity and stability and reduces much working strength of human detection.