A GPU Ray Tracer for Modeling Electromagnetic Scattering from the Human Body

Abstract

A GPU Ray Tracer for Modeling Electromagnetic Scattering from the Human Body Kate Williams, Borja Gonzalez-Valdes, Luis Tirado, Zhongliang Chen, Jose Martinez, Carey Rappaport æPerson-borne weapons and explosives present major security threats at infrastructures such as airports. In many facilities, millimeter-wave screening portals are used to identify anomalous objects hidden underneath peopleÍs clothing. An accurate reconstruction of the human body is required to ensure individuals are not carrying hidden weapons or explosives. There is a demand for improved hardware and software for these systems in order to improve the efficiency and accuracy of the reconstructions. æA ray tracing algorithm has been developed for approximating electromagnetic scattering in the nearfield from a simulated human. æThe algorithm was implemented using the NVIDIA OptiX ray tracing engine. ææConventional full-wave methods used for computing electromagnetic scattering, such as Finite Difference Frequency Domain or the Method of Moments, provide exact solutions but are usually too computationally intensive to be used for 3D simulation and inversion. æRay tracing is inherently parallelizable and has the potential for real-time simulation, especially when taking advantage of graphical processing tools developed for hyper-realistic computer generated imagery. Multiple frequencies can be calculated at no additional cost, unlike full-wave methods. æThe fields computed with the ray tracing algorithm are used to generate accurate SAR reconstructions of the human body. æNumerical results are compared with Method of Moments solutions.