This paper discusses a project initiated by the Department of Engineering and Technology at the Manchester Metropolitan University in cooperation with the Manchester Airport Aviation Services to develop a weapons detection system for airport baggage screening. The weapons detection system uses image recognition techniques to aid operator interpretation of x-ray images. A particular problem of identifying 3D objects from unknown 2D viewpoints has been addressed by using a novel form of a geometric descriptor. Neural networks allow for the creation and mining of large weapon image libraries. Rather than employing a complex mathematical model, the neural network has the simplicity to average features and yet retain the inherent subtle differences which allow for accurate matching.
Weapons detection system for airport baggage screening
Measurement and Control ; 38 , 5 ; 140-146
2005
7 Seiten, 15 Quellen
Aufsatz (Zeitschrift)
Englisch