A simple model is presented of a possible inspection regimen applied to each leg of a cargo containers' journey between its point of origin and destination. Several candidate modalities are proposed to be used at multiple remote locations to act as a pre-screen inspection as the target approaches a perimeter and as the primary inspection modality at the portal. Information from multiple data sets are fused to optimize the costs and performance of a network of such inspection systems. A series of image processing algorithms are presented that automatically process X-ray images of containerized cargo. The goal of this processing is to locate the container in a real time stream of traffic traversing a portal without impeding the flow of commerce. Such processing may facilitate the inclusion of unmanned/unattended inspection systems in such a network. Several samples of the processing applied to data collected from deployed systems are included. Simulated data from a notional cargo inspection system with multiple sensor modalities and advanced data fusion algorithms are also included to show the potential increased detection and throughput performance of such a configuration.

Date Published: 5 October 2007
PDF: 6 pages
Proc. SPIE 6736, Unmanned/Unattended Sensors and Sensor Networks IV, 67360M (5 October 2007); doi: 10.1117/12.738354

Published in SPIE Proceedings Vol. 6736:
Unmanned/Unattended Sensors and Sensor Networks IV
Edward M. Carapezza, Editor(s)