Material Identification From Radiographs Without Energy Resolution

We propose a method for performing material identification from radiographs without energy-resolved measurements. Material has wide variety of applications, including in biomedical imaging, nondestructive testing, and security. While existing techniques radiographic make use dual energy sources, energy-resolving detectors, or additional (e.g., neutron) measurements, such setups are not always practical— requiring hardware complicating imaging. tackle resolution, allowing standard X-ray systems to provide information hardware. Assuming setting where the geometry each object scene is known materials come set possible materials, we pose problem as combinatorial optimization with loss function that accounts presence scatter an unknown gain branch bound algorithm efficiently solve it. present experiments on both synthetic data real, experimental relevance security applications— thick, dense objects imaged MeV X-rays. show can be efficient accurate, example, three shells (two copper, one aluminum), our ran six minutes consumer-level laptop identified correct being among top 10 best matches out 8,000 possibilities.