Super-Resolution of Active Terahertz Imaging via SRGAN

Due to the absence of ionizing radiation, terahertz waves hold great potential for applications in human security inspection. However, accurate recognition of hidden targets from terahertz images poses a significant challenge to its practical applications. This study investigates the feasibility of Generative Adversarial Networks (GANs) for super-resolution processing of active terahertz (THz) images. Through the application of deep learning techniques, we propose an advanced approach to enhance the spatial resolution of active THz imaging by translating low-resolution THz imaging images into high-resolution counterparts. The experimental findings underscore the notable advantages of our method in elevating the quality of THz images with more detailed image information. This research holds significant implications for augmenting the application potential and performance of THz imaging technology, thereby laying a solid foundation for the practical deployment of high-resolution THz imaging in diverse domains, including medical diagnostics and security screenings.