Color Normalization for Robust Automatic Bill of Materials Generation and Visual Inspection of PCBs
Color Normalization for Robust Automatic Bill of Materials Generation and Visual Inspection of PCBs
Wednesday, December 9, 2020: 11:15 AM
Summary:
A Bill of Materials (BoM), is the list of all components present on a Printed Circuit Board (PCB). BoMs are useful for multiple forms of failure analysis and hardware assurance. In this paper, we build upon previous work and present an updated framework for automatically extracting a BoM from optical images in order to keep up-to-date with technological advancements. This is accomplished by revising the framework to emphasize the role of machine learning and incorporate domain knowledge of PCB design and hardware trojans. In implementing the first few stages of this framework, we explore the effect of color profile normalization on detection and classification algorithm accuracy. This is accomplished by collecting PCB images under a wide variety of illumination conditions and testing the performance of a suite of algorithms before and after normalization. This colorspace noise reduction is highly desirable for machine learning methods, as it allows for higher accuracy and reduces the number of ground truth images necessary for training. Finally, we pose methods to improve future iterations of image normalization and color profile extraction.
A Bill of Materials (BoM), is the list of all components present on a Printed Circuit Board (PCB). BoMs are useful for multiple forms of failure analysis and hardware assurance. In this paper, we build upon previous work and present an updated framework for automatically extracting a BoM from optical images in order to keep up-to-date with technological advancements. This is accomplished by revising the framework to emphasize the role of machine learning and incorporate domain knowledge of PCB design and hardware trojans. In implementing the first few stages of this framework, we explore the effect of color profile normalization on detection and classification algorithm accuracy. This is accomplished by collecting PCB images under a wide variety of illumination conditions and testing the performance of a suite of algorithms before and after normalization. This colorspace noise reduction is highly desirable for machine learning methods, as it allows for higher accuracy and reduces the number of ground truth images necessary for training. Finally, we pose methods to improve future iterations of image normalization and color profile extraction.
See more of: Hardware Attacks, Security, and Reverse Engineering
See more of: EDFAS Virtual Workshop
See more of: EDFAS Virtual Workshop