Rising vaccine production and complex visual characteristics of freeze-dried products have highlighted a critical need for accurate, high-speed automated quality control. Current inspection procedures, that rely on human vision or line cameras, have undesirable error rates. We propose a novel use of polarimetric imaging for defect capture and compare the performance of polarimetric imaging to RGB imaging for defect detection on vaccine vials with freeze-dried product. Vaccine vials with artificial defects (scratches and fibers) and without defects but with product appearance variations (streaks) are prepared. We capture a data set of RGB images and polarimetric images: Polarization Intensity (PI), Degree of Linear Polarization (DoLP), Angle of Polarization (AoP). We find that the differences between product variation and defects in RGB images are not statistically significant with α = 0.01 (t(8) = 2.088 for scratch vs. streak, t(8) = 2.789 for fiber vs. streak). In contrast, the differences between product variation and defects for polarimetric imaging are statistically significant for all polarization characteristics with α = 0.01 (PI: t(8) = 39.753 for scratch vs. streak, t(8) = 13.039 fiber vs. streak, DoLP: t(8) = 16.537 for scratch vs. streak, t(8) = 17.018 for fiber and streak, AoP: t(8) = 6.764 for scratch vs. streak, t(8) = 4.702 for fiber vs. streak). This indicates that polarimetric imaging may be used as a more effective technique than RGB imaging for defect detection.
Automating visual inspection of vials containing freeze-dried products is a difficult problem due to the complex appearance of freeze-dried materials. To existing inspection equipment, defects and typical product appearance variation often appear very similar. We contend that these shortcomings necessitate a new approach and propose a multimodal sensor integration framework. We study polarimetric imaging combined with edge detection to improve the detection of defects such as scratches in glass vials. Stokes vectors enable the extraction and display of different polarization characteristics. Edge detection compares the effectiveness of RGB and polarimetric imaging for defect detection. A novel mathematical description of the outline strength decreases edge detection complexity. Our work shows that polarimetry, in combination with edge detection, enhances the detection of defects. We find that a combination of polarization characteristics in the hue-saturation-value (HSV) false coloring scheme outperforms all individual polarization characteristics and RGB imaging, achieving a 100% detection accuracy at an outline strength of 0.7. Image resolution impacts the effectiveness of edge detection as well. A resolution of 1500 x 1000 pixel works best for defect detection in our dataset. Defect detection is most sensitive to scratch depth, while scratch length and thickness are less important. Our findings not only apply to defect detection on glass vaccine vials, but also to the inspection of other reflective surfaces such as plastics and metals. This work can be expanded to problems that require material detection. Denise Tellbach is a graduate student in Mechanical Engineering at MIT. Her research interests are in IoT, sensing, and applying machine learning to enhance the functionality of industrial quality control. Denise received a double master’s degree in Management Science and Mechanical Engineering from RWTH Aachen University and Tsinghua University in 2018 and her undergraduate degree from RWTH Aachen University in 2016. In the past, she has developed a maturity model for the digitalization of production control, she has worked on cyber-physical systems modeling and on reliability assessment focusing on the electric grid. She joined the AutoID Lab as a graduate student under Prof. Sarma in 2019 and is a Presidential Fellow at MIT (2019). Praveer Sharan is a researcher affiliated with the Auto ID Lab at the Massachusetts Institute of Technology. His research interests are in data analysis, unmet needs in healthcare, edge detection, and machine learning to boost efficiency of healthcare. Praveer is scheduled to graduate in 2022 and is applying for computer science programs. In the past, he has conducted a review on the ability of neural networks to automatically discriminate between healthy versus unhealthy coughs based on the processed audio of the cough. He has worked on research projects and conducted research with the MIT computer science department for the last two summers.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.