A Study on 1D-Point Cloud Using Through-Wall mm Wave Radar for Human Pose Estimation
Kritthee Phaisiri, Chainarong Kittiyanpunya, Nattakarn Shutimarrungson, Akkarat Boonpoonga, Prayoot Akkaraekthalin, Kamol Kaemarungsi · 2024
This paper presents human pose estimation using millimeter-wave (mmWave) radar technology capable of penetrating obstacles. Operating within the frequency modulated continuous wave (FMCW) at 60 - 64 GHz, our radar system generates signals conducive to accurate human pose representation. Our data processing framework focuses on producing point cloud data, effectively capturing each human movement. Results are visualized through a 2D graph, with the z-axis representing time. Experimental data, collected from various activities conducted behind different walls, underwent thorough analysis, comparing point cloud positions with actual human postures. The outcomes affirm the effectiveness of mmWave radar systems, positioned behind obstacles, in accurately extracting human postures.