Localization and tracking of radioactive source carriers in person streams

The localization and tracking of radioactive sources in public facilities like airports or stations is a problem of highest security relevance. The accumulation and the severity of terrorist attacks during the past decade give reason to the assumption that future attacks could also involve radioactive material packaged with conventional explosives. The only way to avoid such kind of attacks is to localize and arrest the person carrying the material to its destination. But since radiation is not perceivable by human beings, the security guards are largely dependent on technical decision support to perform this task. We consider a security assistance system comprising three gamma scintillation detectors that are distributed along a corridor wall to check passing people for radioactive material. Furthermore, the system consists of a set of tracking sensors simultaneously providing the positions of all persons during their walk through the corridor. In this paper we propose techniques to estimate the assignment of radioactive detections to person tracks. These techniques provide a measure for each person that reflects the probability that the person is a radioactive source carrier. The problem of source localization is thus reduced to a matching problem between person tracks and sequences of count rates.