Printed Circuit Board Defect Detection Using Deep Learning via A Skip-Connected Convolutional Autoencoder - PubMed Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2021 Jul 21;21(15):4968.
doi: 10.3390/s21154968.

Printed Circuit Board Defect Detection Using Deep Learning via A Skip-Connected Convolutional Autoencoder

Jungsuk Kim  1 , Jungbeom Ko  1 , Hojong Choi  2 , Hyunchul Kim  3
Affiliations

Printed Circuit Board Defect Detection Using Deep Learning via A Skip-Connected Convolutional Autoencoder

Jungsuk Kim et al. Sensors (Basel). .

Abstract

As technology evolves, more components are integrated into printed circuit boards (PCBs) and the PCB layout increases. Because small defects on signal trace can cause significant damage to the system, PCB surface inspection is one of the most important quality control processes. Owing to the limitations of manual inspection, significant efforts have been made to automate the inspection by utilizing high resolution CCD or CMOS sensors. Despite the advanced sensor technology, setting the pass/fail criteria based on small failure samples has always been challenging in traditional machine vision approaches. To overcome these problems, we propose an advanced PCB inspection system based on a skip-connected convolutional autoencoder. The deep autoencoder model was trained to decode the original non-defect images from the defect images. The decoded images were then compared with the input image to identify the defect location. To overcome the small and imbalanced dataset in the early manufacturing stage, we applied appropriate image augmentation to improve the model training performance. The experimental results reveal that a simple unsupervised autoencoder model delivers promising performance, with a detection rate of up to 98% and a false pass rate below 1.7% for the test data, containing 3900 defect and non-defect images.

Keywords: PCB defeat detection; autoencoder; deep learning; detect detection; printed circuit board manufacturing.

PubMed Disclaimer

Conflict of interest statement

The authors declare no conflict of interest.

Figures

Figure 1
Figure 1
The 400×400 PCB images are cropped from Huang and Wei’s PCB dataset [12]. The number below each PCB image is the name of the reference PCB. The numbers below the cropped images are the name of the reference image files.
Figure 2
Figure 2
Examples of PCB defects.
Figure 3
Figure 3
System overview of PCB defect detection (To enhance the quality of the data, we applied preprocessing to the PCB dataset. The quantity of the data for training is fulfilled by the data augmentation step, and all the data are inputted into our proposed autoencoder model. After the training, the trained model generates a non-defect image from the defect image, and image subtraction between these two images enables us to find the exact defect shape and location).
Figure 4
Figure 4
Working process of an autoencoder: transforming input data to compressed latent vectors and then decoding it as the data.
Figure 5
Figure 5
Architectures of the skip-connected convolutional autoencoder and convolutional autoencoder. The arrows indicate the skip connections.
Figure 6
Figure 6
Training losses of the two models. The green line graph is the training loss of the skip-connected convolutional autoencoder, whereas the blue one indicates that of the convolutional autoencoder. Both models show decreased training loss according to the learning rate.
Figure 7
Figure 7
Results from the six test images. The subtraction result is the difference between the generated image and the input image. The binary dilation result is an image whose noise is removed by using the dilation technique.
Figure 8
Figure 8
Results from the new PCB images. The defects of input images were artificially drawn.
See this image and copyright information in PMC

Similar articles

See all similar articles

Cited by

See all "Cited by" articles

References

    1. Radiant Vision System. [(accessed on 19 July 2021)]; Available online: https://www.radiantvisionsystems.com/products.
    1. Guo F., Guan S.-A. Research of the Machine Vision Based PCB Defect Inspection System; Proceedings of the International Conference on Intelligence Science and Information Engineering; Washington, DC, USA. 20–21 August 2011; pp. 472–475.
    1. Koch J., Gritsch A., Reinhart G. Process design for the management of changes in manufacturing: Toward a Manufacturing Change Management process. CIRP J. Manuf. Sci. Technol. 2016;14:10–19. doi: 10.1016/j.cirpj.2016.04.010. - DOI
    1. Anoop K.P., Kumar S. A Review of PCB Defect Detection Using Image Processing. Intern. J. Eng. Innov. Technol. 2015;4:188–192.
    1. Park J.-K., Kwon B.-K., Park J.-H., Kang D.-J. Machine learning-based imaging system for surface defect inspection. Int. J. Precis. Eng. Manuf. Technol. 2016;3:303–310. doi: 10.1007/s40684-016-0039-x. - DOI