A novel Markov random field-based clustering algorithm to detect high-Z objects with cosmic rays

We have developed a novel algorithm based on Markov random fields capable of detecting special nuclear material in large-scale volumes such as cargo containers, using cosmic muons as probes. Since the amount of muon scattering is dependent on the Z² of the material traversed, strong scattering in a localized area is indicative of high-Z material being present. For scanning purposes in freight harbors and similar, a decision should be made in 1 minute. We will show that our algorithm is capable of identifying a solitary uranium block in a cargo container with 100% accuracy in 1 minute, and suffers only a minor degradation of performance for the same block shielded by a large rock volume.