When
12:30 – 1:30 p.m., March 26, 2026
Where
Speaker: Jacob Rezac, National Institute of Standards and Technology
Title: Binary Segmentation with Precomputed Filters applied to X-Ray Computed Tomography Scans of Semiconductors
Abstract: Many problems in science and engineering involve a comparison of images collected through a physical measurement. For example, X-Ray Computed Tomography (CT) can be used to track the degradation of an object over time. Object segmentation is critical for this type of defect detection, typically done by with an error-prone multistep procedure that includes object measurement, solving of an inverse problem, and finally segmentation. We instead solve the inverse problem directly for a segmentation of the object interior, using a prior collection of possible shape solutions based on previous experiments or scientific knowledge. We explain this procedure with an application involving X-Ray CT scans of semiconductors.
Abstract: Many problems in science and engineering involve a comparison of images collected through a physical measurement. For example, X-Ray Computed Tomography (CT) can be used to track the degradation of an object over time. Object segmentation is critical for this type of defect detection, typically done by with an error-prone multistep procedure that includes object measurement, solving of an inverse problem, and finally segmentation. We instead solve the inverse problem directly for a segmentation of the object interior, using a prior collection of possible shape solutions based on previous experiments or scientific knowledge. We explain this procedure with an application involving X-Ray CT scans of semiconductors.