Tongxin xuebao (Jul 2024)
Multi-target vital sign estimation based on sparse representation under complex scenes
Abstract
Focusing on the issue that millimeter-wave radar was difficult to accurately estimate the vital signs of multiple moving targets in complex indoor scenes, a multi-target vital sign estimation method based on sparse representation under complex scenes was proposed. Firstly, the echo data was preprocessed to acquire the point clouds of target and background. After that, a dynamic clutter suppression model was constructed to filter out the dynamic interference. In what follows, the echo data was assigned to the corresponding target, and multi-target tracking could be achieved by exploiting the extended Kalman filter to extract the phase information of the chests of the multi-moving targets. Subsequently, with the sparsity of respiratory and heartbeat signals in the frequency domain, a data-driven adaptive dictionary construction method was proposed to effectively separate respiratory and heartbeat signals. Finally, high precision multi-target vital signs estimation could be achieved by using the sparse reconstruction method. Amount of experimental results in the actual scenes show that the proposed method can effectively perceive the vital signs of multi-target in complex dynamic clutter scenes as compared to the state-of-the-art vital sign estimation methods.
