Advanced container inspection system based on dual-angle X-ray imaging method

Abstract

In 2007, the U.S. Congress mandated the implementation of the “Security and Accountability For Every Port Act of 2006,” which requires complete scanning of 100% of U.S.-bound shipping containers. To address this requirement, we developed a container inspection method that enables continuous high-speed screening, with considerable performance improvement. In this study, we developed a fixed-type high-precision container inspection system using dual-angle X-ray beams from a 9 MV linear accelerator (LINAC). We first calculated the X-ray irradiation angle-dependent changes in the contrast-to-noise ratio (CNR) of the images via Monte Carlo simulation. Using the calculated CNRs, the primary and secondary angles of the X-ray beam were set to 0° and 2.8°, respectively. A system based on the proposed dual-angle X-ray imaging technology was installed and evaluated by scanning a real cargo container truck. For the evaluation, we designed test equipment based on the ANSI N42.46 report and examined the beam penetration power, contrast sensitivity, spatial resolution, and wire detectability of the developed system. The maximum penetration thicknesses for the primary and secondary angle beams were found to be 410 and 400 mm, respectively. At the primary beam angle, the contrast sensitivities were 1.52% and 0.49% when the thicknesses of the steel plate were 80% and 50% of the maximum penetration thickness, respectively. At the secondary angle, the sensitivities were 1.88% at 80% maximum penetration thickness, and 0.5% at 50%. A line pattern formed by individual slits of 4.6 mm width could be easily recognized in an acquired image. In addition, the developed system could clearly identify a 1.6 mm diameter copper wire. Further, when a steel plate was added, the change in the wire-recognition ability of the imaging system was found to be similar at both beam angles. These results indicate that the developed system is suitable for container screening using a 9 MV LINAC. Shapes that could not be identified from one beam irradiation angle could more accurately be analyzed using images from two different angles.