Printed Circuit Board BGA Joint Defect Automatic Detection And Recognition Based On X-ray

Aimed at the special requirements of the BGA solder joint defect detection in X-ray printed circuit board, this paper studies the defect detection and recognition technology based on X-ray real time imaging.In the field of printed circuit board detecting, the defects, such as blowholes, solder bridge and poor soldering, appear as the solder joints under the BGA package devices are not visible, which can affect the quality and security of PCB products. So BGA solder joint image defect detection is very important. The traditional method of X-ray image detecting is affected by technical quality and experience of comment piece staff. Film method has some drawbacks such as low detection efficiency and complicated operation, which are also difficult to save film information and automate the testing process. The online detection and analysis is objective, normative and standard.This technology can overcome misjudgment by manual effectively if used X-ray real time detecting system.According to detective theory of ray imaging system, this paper establishes an high resolution X-ray real time detecting system on BGA solder joint defect, which can test and evaluate the nature of defect automatically, and then evaluate the solder joint quality.There are some drawbacks such as random noise and low brightness in X-ray solder joint detecting image. Therefore we use noise reduction method of frame integral and superposition in this paper. This method can eliminate the random noise effectively and retain the original image information maximumly. Image enhancement technology is used to enhance the reality effect. It lays a good foundation for automatic extraction of defect characteristics. Classify the defect according to the type of BGA Solder Joint defects. Extract the feature parameters such as the number of connected domains, regional center of gravity, circumference and area of connected domains, local information entropy ratio and so on. And then build classifier to determine feature selection using the method of statistical pattern recognition.Finally statistical recognition method is applied to identify the Ray BGA solder joint defecting image. Set classifier through a combination of theory and prior knowledge and input the experimental data to make the classifier obtain training results. And then correct the results of classifier training continuously until a satisfactory classification results is obtained. The experiment shows that statistical identification method can identify defects effectively which also have a higher accuracy and the algorithm are more efficient.