A fast non-destructive defect detection method for complex PCB surfaces

In the manufacturing process of a ceramic resonant cavity with a PCB, different components of the PCB have different light reflectivity, which greatly reduce the defect detection accuracy. In this paper, a defect detection system with its defect detection algorithm for complex PCB surface is proposed, which is suitable for manufacturing workshop. Some images of good samples are selected in advance and are fused as the template image. Then the template image is divided into three parts due to the light reflectivity of different material bases. Through the template matching between the template image and the image to be detected, an affine transformation matrix is obtained. After matching, the difference between the template and the target image can greatly reduce the interference caused by the different light reflectivity of the material base, and improve the detection efficiency of cross-regional defects. The algorithm has been well applied in actual production. At present, the number of samples detected has exceeded 100,000, and the average detection speed is about 1s/piece. Using the result of manual inspection as standard, every actual defective sample is selected, and the precision rate of samples is 98.3%.