Deep learning based solder joint defect detection on industrial printed circuit board X-ray images

Qianru Zhang,Chau Yuen,Hirunima Jayasekara,Meng Zhang,Chinthaka Gamanayake,Jenny Low,Yong Liang Guan,Xiang Liu,Chia-Wei Woo,Zehao Geng

doi:10.1007/s40747-021-00600-w

Abstract

With the improvement of electronic circuit production methods, such as reduction of component size and the increase of component density, the risk of defects is increasing in the production line. Many techniques have been incorporated to check for failed solder joints, such as X-ray imaging, optical imaging and thermal imaging, among which X-ray imaging can inspect external and internal defects. However, some advanced algorithms are not accurate enough to meet the requirements of quality control. A lot of manual inspection is required that increases the specialist workload. In addition, automatic X-ray inspection could produce incorrect region of interests that deteriorates the defect detection. The high-dimensionality of X-ray images and changes in image size also pose challenges to detection algorithms. Recently, the latest advances in deep learning provide inspiration for image-based tasks and are competitive with human level. In this work, deep learning is introduced in the inspection for quality control. Four joint defect detection models based on artificial intelligence are proposed and compared. The noisy ROI and the change of image dimension problems are addressed. The effectiveness of the proposed models is verified by experiments on real-world 3D X-ray dataset, which saves the specialist inspection workload greatly.

Highlights

With the improvement of electronic circuit production methods, the production speed of electronic products, such as mobile phones and notebook computers, is increasing
Imaging-based automatic inspection methods such as optical imaging and thermal imaging are often used in the production line for quality control [3,4,5,6]
There are some deep learning-based methods implemented in printed circuit board (PCB) defect detection as mentioned, few researches focus on X-ray imaging and addressing the varying number of slices

Summary

Introduction

With the improvement of electronic circuit production methods, the production speed of electronic products, such as mobile phones and notebook computers, is increasing. Sometimes the machine inspection is not accurate, which sends a large number of normal solder joints to the specialists that increases the manual workload. The main problem is to detect true defect solder joints from misclassified dataset and to reduce the manual workload. There are some deep learning-based methods implemented in PCB defect detection as mentioned, few researches focus on X-ray imaging and addressing the varying number of slices. The dedicatedly designed pre-processing methods can address the varying number of slices problem and the incorrect ROI problem. Since few deep learning-based methods address the varying numbers of slices problem and the incorrect ROI problem in PCB X-ray imaging, the performance is compared among each other, which can help with the researches with similar problems. Experiment and performance results are presented in Section “Experiment and results” and we conclude the work in Section “Conclusion”

Methodology

Results

Conclusion