Terahertz (THz) waves offer transmission and high-resolution capabilities, enabling physical information such as subsurface structure, dielectric properties, and molecular-scale dynamics that conventional visible cameras cannot readily acquire.
By integrating THz devices and measurement techniques with physical models and AI, we develop next-generation THz sensing infrastructure for practical deployment.
Material Property Extraction Using Physics-Informed Neural Networks
We incorporate electromagnetic phenomena such as propagation, scattering, and absorption of terahertz waves into AI models to develop PINN-based methods for extracting material properties. Amplitude and phase information in THz spatiotemporal signals are used to estimate dielectric properties accurately, enabling stable and practical THz sensing analysis.
AI-Based Fusion of Terahertz and Visible Information
We develop AI methods that fuse physical information from terahertz waves, including transmission and dielectric characteristics, with shape, colour, and texture information contained in visible images. By integrating complementary sensor information, we aim to establish a multimodal sensing platform for reliable and accurate recognition and measurement.