Moving Human Detection Method in Through-the-wall Radar Based on Robust Principal Component Analysis and YOLOv8

‍ ‍With the rapid advancement of urbanization， building structures have become increasingly dense， resulting in more complex urban environments. In such settings， dangerous individuals often hide in shielded spaces obstructed by buildings. Therefore， it is urgently necessary to develop effective detection methods for shielded areas. Through-the-wall radar technology， which emits electromagnetic waves capable of penetrating obstacles such as walls， offers a promising solution for detecting moving humans in these environments. However， through-the-wall radar detection is challenged by strong wall clutter and low-quality radar imagery. This study proposes a through-the-wall radar moving human detection method based on robust principal component analysis （RPCA） and YOLOv8 to address these issues. The aim is to enhance the accuracy and reliability of detecting moving humans in shielded environments. In the proposed method， RPCA is utilized to extract moving human echoes from the original radar returns of each channel， leveraging the low-rank nature of wall clutter and the sparsity of moving target echoes. Following this， the time-domain back-projection algorithm is employed to generate radar images. The Phase Coherence Factor （PCF） weighting method is applied to enhance image quality， based on the phase correlation between target and sidelobe positions across radar transceiver channels. This technique effectively suppresses azimuthal sidelobes. Finally， the enhanced radar images are processed using the deep neural network YOLOv8 to detect moving humans. The effectiveness and robustness of the proposed method were verified through both numerical simulations and field experiments. In the simulations， a scenario with two moving individuals behind a wall was designed. The results showed that RPCA effectively suppressed static clutter such as wall reflections and successfully extracted the echoes of the two moving individuals. The proposed approach significantly improved the signal-to-clutter ratio （SCR） compared with other methods. The experimental results showed that the YOLOv8-based moving person detection model could successfully detect the number and positions of moving persons behind the wall. In field experiments， a MIMO radar system was deployed to detect multiple moving targets behind a 20 m wall. The results demonstrated that the proposed method could accurately detect the number and locations of moving persons while effectively suppressing clutter. In conclusion， the through-the-wall radar moving human detection method based on RPCA and YOLOv8 provides a reliable and efficient solution for detecting moving humans in shielded spaces. It effectively overcomes the challenges of strong wall clutter and low-quality radar images， making it a valuable tool for applications such as security surveillance， search and rescue operations， and law enforcement in complex urban environments. Future research may focus on optimizing computational efficiency and expanding the method’s applicability to a wider range of scenarios.