Image processing techniques for metallic object detection with millimetre-wave images

In this paper we present a system for the automatic detection and tracking of metallic objects concealed on moving people in sequences of millimetre-wave images, which can penetrate clothing, plastics and fabrics. The subjects are required to enter one at a time and turn round slowly to ensure complete coverage for the scan. The system employs two distinct stages: detection and tracking. In this paper a single detector, for metallic objects, is presented which utilises a statistical model also developed in this paper. Target tracking is performed using a Particle Filter. Results are presented on real millimetre-wave image test sequences and indicate an excellent rate of success for threat identification. Encouraging results for target tracking are also reported.