Automated Trace and Copper Plane Extraction of X-ray Tomography Imaged PCBs

Reverse engineering (RE) of Printed circuit boards (PCBs) can be achieved through X-ray Computed Tomography (CT) in a non-destructive manner. For practical applications such as obsolescence replacement and hardware assurance, it is important to perform RE as automated and fast as possible. While our past work has focused on automated via detection, this paper introduces a framework for automated and unsupervised extraction of traces and copper conducting planes in an X-ray CT imaged PCB. We employ a series of algorithms from image processing, computer vision, and graph theory for a robust solution that produces a high fidelity result. We compare our results against off-the-shelf methodologies utilizing ridge detection and Canny edge detection in combination with connected components and show the necessity of utilizing a graph-theoretic approach. Our results are demonstrated on an in-house designed PCB for qualitative and quantitative analysis. On average across all layers, our proposed framework outperforms the relevant methods with IoU, SSIM, Correlation, and Dice image quality scores of 0.9166, 0.9954, 0.8952, and 0.9453, respectively.