Study on Scattering Correction of the 60Co Gantry-Movable Dual-Projection Digital Radiography Inspection System

Nuclear safety and security is getting more and more attention all over the world. Vehicles in and out of nuclear facilities need to be strictly inspected to prevent carrying of nuclear materials, explosives or other dangerous items. A new-type vehicle inspection system — 60Co gantry-movable dual-projection digital radiography inspection system is developed in Tsinghua University under the support of China’s Nuclear Energy Development Project, which uses two 60Co as radiation sources. The radiation sources are arranged at the bottom and side of the vehicles to be detected and the ionization chamber detectors are set in the two side of the gantry correspondingly. With moving of the sources and gantry synchronously, the system can obtain both side-view and upward-view images of the vehicles simultaneously [1]. However, there is a problem that the scattered rays from source of one projection can enter the detector array of another projection to form an interference signal. Experiments show that this kind of scattering noise can account for 20%, resulting in blurring and even artifacts, especially in thicker areas of the image. This problem needs to be solved urgently. According to the characteristics of the Compton scattering effect, it is inferred that there is a certain non-linear mapping relationship between the scattering distribution of the detector array of one projection and the mass distribution in the plane of the other projection. This paper attempts to use BP neural network to learn this mapping relationship to quantitatively remove this kind of scattering noise. The results show that this method has certain effects on the removal of artifacts and blur caused by scattering. This method has the advantage of being fast and more targeted, compared with traditional image post-processing methods.