Noise and Artifact Reduction in Computed Tomography (CT) Images

Using experimentally-verified simulation we rank the most troublesome artifact mechanisms in high-energy 2D industrial computed tomography (CT) reconstruction. These artifacts, which include imprecise geometry, beam-hardening, sampling issues, scatter, and non-linearities, are fundamentally different from their appearance in medical and other low-energy CT implementations. We will demonstrate several new and original artifact reduction algorithms targeted at the most critical artifacts, those which hamper detection and mensuration of anomalies. We propose nine promising candidate algorithms, of which six are unique to high-energy x-ray (i.e., > 5 MeV) applications. In Phase I we will evaluate and deliver demonstrations of the most promising algorithms. The Phase II program completes all demonstrations and design, and delivers a prototype artifact reduction package incorporating all successful artifact reduction algorithms. This package will be designed for retrofit of existing HECT2 and ICT1500/2500 systems and for new high-energy CT designs. JDLL team members have accumulated unmatched depth and expertise in high-energy rocket motor CT. Our team includes project scientists/engineers from original HECT for Trident, from ICT data analysis and data structuring, and the HECT2 upgrade for Minuteman. JDLL's artifact reduction innovations have reduced imaging artifacts in HECT2 and numerous other military CT systems.