RPCA-Based High Resolution Through-the-Wall Human Motion Feature Extraction and Classification

Radar-based assisted living has received significant amount of research interest in recent years. However, most of the existing works are focused only on free space detection. When it comes to the through-the-wall human motion detection, the wall media and indoor static non-human targets would cause clutters and significantly corrupt the motion information of human targets behind wall. In this work, we first propose to use a low center-frequency ultra-wideband (UWB) radar system to probe the behind wall scene. Then, a Robust Principal Component Analysis (RPCA) based subspace decomposition technique is employed not only to remove the stationary clutters in raw range slow-time map but also to mitigate the multipath effects in the time-frequency map. Onsite experiments of human motions detection behind a single layer of reinforced concrete wall are carried out to investigate the performance of the technique. Lastly, a two-dimensional (2D) PCA based method and a convolutional neural network (CNN) based method are employed for the motion classification. The results based on the enhanced features obtained by the proposed method show a higher accuracy than those based on the traditional features through mean value subtraction.