Coffee Bean Characterization Using Terahertz Sensing

Coffee bean sorting is currently based primarily on visual appearance and near-infrared techniques that probe the bean's skin. However, sorting based on compositional differences has significant potential to optimize the roasting process. We present a novel coffee bean sorting method using terahertz (THz) spectroscopy, which effectively penetrates both green and roasted beans. Our findings show that the optical properties of coffee beans at THz frequencies are primarily governed by internal moisture levels. To demonstrate industrial feasibility, we implement a robot-guided THz sensing system capable of scanning beds of beans for automated sorting. More broadly, our results confirm the potential of THz technology for moisture content analysis across various applications.