IMAGE ANALYSIS ALGORITHMS FOR DUAL MODE IMAGING SYSTEMS

Conference · Fri Jun 11 04:00:00 EDT 2010

OSTI ID:1004529

Robinson, Sean M; Jarman, Kenneth D; Miller, Erin A; Misner, Alex C; Myjak, Mitchell J; Pitts, W Karl; Seifert, Allen; Seifert, Carolyn E; Woodring, Mitchell L

The level of detail discernable in imaging techniques has generally excluded them from consideration as verification tools in inspection regimes where information barriers are mandatory. However, if a balance can be struck between sufficient information barriers and feature extraction to verify or identify objects of interest, imaging may significantly advance verification efforts. This paper describes the development of combined active (conventional) radiography and passive (auto) radiography techniques for imaging sensitive items assuming that comparison images cannot be furnished. Three image analysis algorithms are presented, each of which reduces full image information to non-sensitive feature information and ultimately is intended to provide only a yes/no response verifying features present in the image. These algorithms are evaluated on both their technical performance in image analysis and their application with or without an explicitly constructed information barrier. The first algorithm reduces images to non-invertible pixel intensity histograms, retaining only summary information about the image that can be used in template comparisons. This one-way transform is sufficient to discriminate between different image structures (in terms of area and density) without revealing unnecessary specificity. The second algorithm estimates the attenuation cross-section of objects of known shape based on transition characteristics around the edge of the object’s image. The third algorithm compares the radiography image with the passive image to discriminate dense, radioactive material from point sources or inactive dense material. By comparing two images and reporting only a single statistic from the combination thereof, this algorithm can operate entirely behind an information barrier stage. Together with knowledge of the radiography system, the use of these algorithms in combination can be used to improve verification capability to inspection regimes and improve and substantially increase the likelihood of detecting material diversion.

• Research Organization: Pacific Northwest National Laboratory (PNNL), Richland, WA (US)
• Sponsoring Organization: USDOE
• DOE Contract Number: AC05-76RL01830
• OSTI ID: 1004529
• Report Number(s): PNNL-SA-73315
• Country of Publication: United States
• Language: English

Related Subjects

11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS
ALGORITHMS
ATTENUATION
NUCLEAR MATERIALS MANAGEMENT
PERFORMANCE
POINT SOURCES
RADIOACTIVE MATERIALS
SHAPE
SPECIFICITY
STATISTICS
VERIFICATION
arms control
dismantlement
imaging
information barrier
verification