Aiming at the defects of FPC solder surface,a detection method based on image processing is proposed. Firstly,the average image of no defect FPC is obtained as a reference,and defect’s location can be achieved by image difference method. And then,information entropy is introduced to describe color fluctuation and geometrical morphol-ogy on FPC solder for quantifying and extracting texture feature. The color and area feature can also be used to dis-criminate non-defective and defective solder. So,all the defective solder can be detected according to color,area and texture characteristics. The performance of the proposed defect detection algorithm is finally evaluated on-line testing. The average detection time of 50 solders is 300ms,and the detection accuracy can reach up to 97. 8%. It is very fit for real time detection.%针对FPC焊盘表面缺陷,提出一种基于图像处理技术的智能检测方法。文章首先获取无缺陷FPC的平均图像为参考,并采用图像差影法对缺陷定位。接着,引入信息熵概念对FPC焊盘表面的几何形貌、颜色波动性进行量化,实现对纹理特征的量化与提取。通过实验观测得到:缺陷焊盘与无缺陷焊盘在颜色、面积特征上具有明显的区别。因此,检测系统通过焊盘的颜色、面积、纹理特征实现对缺陷的检测。检测准确率高达97.8%,50个焊盘的平均检测时间为300ms。满足在线检测的要求。
展开▼