UC Berkeley

Nuclear safeguards inspections aim to verify nuclear materials at facilities around the world, and are faced with constraints such as a limited amount of time to perform measurements, ideally with minimal disruption to facility operation. Gamma-ray imaging is a promising technology for performing these inspections because it presents an efficient and non-destructive method of quantifying nuclear material. Recent advances in gamma-ray imaging have included integrating mobile radiation imagers with real-time tracking and scene reconstruction algorithms, enabling a mobile mode of operation and 3D localization of gamma-ray sources. This technique, dubbed scene data fusion (SDF), holds promise for facilitating safeguards inspections. While developed for free-moving measurements, SDF also has applications in cases that require static measurements for increased counting statistics. In these applications, the scene information can be used to improve quantitative imaging results. This dissertation employs computer vision techniques with the scene information collected by SDF-enabled imagers to identify objects of radiological interest. Computational methods are developed to determine optimal measurement positions to reduce measurement time while still capturing enough data to quantify nuclear materials within desired statistical uncertainties. The computation techniques are experimentally validated with a series of example models and a safeguards-relevant measurement. The result is a software tool that could enable safeguards inspectors to rapidly formulate a measurement plan in the field and provide a quantitative measurement of materials present with low levels of uncertainty.