Research on Human Posture Recognition Technology Based on Millimeter-wave Radar

Human posture recognition is a highly focused area of research in computer vision and wireless communications. This technology has extensive potential applications, ranging from smart homes and interactive games to medical surveillance and even security and crime deterrence. With ongoing technological advancements, millimeter wave radar has gained prominence in human posture recognition, owing to its penetration capabilities, independence from lighting conditions, and ability to capture minute movements. This research conducted experiments using the AWR1642BOOST Frequency-Modulated Continuous-Wave (FMCW) millimeter-wave radar developed by Texas Instruments. Millimeter wave radar gathered raw data on different human postures, including raising the forearm front raises and lateral raise. Post-processing of the collected raw data was done with MATLAB, yielding features like range-time graphs, micro-Doppler-time graphs, and rhythm velocity for various postures. The method of statistical feature extraction was applied to the micro-Doppler-time graph data of two volunteers lifting their forearms, achieving an impressive recognition rate of up to 99%.