Optimized acquisition geometry for X-ray inspection

X-ray inspection of luggage and cargo containers is nowadays fundamental to guarantee the security in the field of transportation. The simplest inspection systems currently in use are based on transmission radiography, but high absorption objects may hide to the inspector the presence of weapons or other dangerous materials. Dual-energy views are used to automatically recognize nuclear and explosive materials, but also in this case high absorption objects may make the recognition results unreliable. Computerized tomography (CT) partially solves these problems, but it generally represents a costly and slow inspection method. In this paper, we suggest to use a fast, adaptive multi-view system where the geometry of the X-ray projecting cones are set according to an empirical criterion aimed at facilitating the recognition of all the objects inside the scanned volume. Experimental results on simulated and real data are reported in the paper to demonstrate the reasonableness and efficacy of the proposed criterion.