疫情当下，最基本的呼吸道诊断方式是肺部听诊。然而由于听诊时医生与患者不可避免地发生近距离接触，导致医务人员感染的概率升高。如何保障一线医务人员的健康安全，成为国家在当下甚至未来面对类似疫情时的重大难题。为了突破这一困境，本项目立足于非接触式生命体征探测技术，瞄准肺音振动这种微弱而又富含信息的生理运动，拟将超分辨理论、模态分解与线性调频连续波雷达探测相融合，提出非接触式肺音振动检测方法。项目基于肺音音频特征建立呼吸驱动下的肺音振动模型，通过收发天线的布阵优化达到对肺部各个区域波束形成的全局最优设计，基于原子范数最小化实现对肺音振动的超分辨感知，同时提出经验模态分解与变分模态分解相融合的肺音振动分析与识别框架，解决呼吸和心跳强干扰情况下宽带微弱振动信号的测量问题。本项目提出了新的非接触式生命体征探测研究方向，研究成果将为我国医疗卫生领域诊断与检测技术的发展提供重要的研究积累和技术支撑。

In the current epidemic, the first way of diagnosis of respiratory symptoms is auscultation of lung sounds. However, due to the inevitable close contact between the doctor and the patient during auscultation, the probability of infection of medical staff increases. How to protect the medical staff in such a situation has become a significant worldwide problem. In order to get out of this situation, we propose to detect the lung abnormality remotely, by fusing super-resolution theory and mode decomposition with frequency modulated continuous wave radar, based on the features of the weak lung vibration. Firstly, a lung vibration model driven by breathing and corresponding to the characteristics of lung sound audio is built. Through the optimization of the transmitting and receiving arrays, the global optimal design for beamforming towards various regions of the lung is achieved. A framework for detecting and identifying the lung vibration is designed by fusing empirical and variational mode decomposition methods, in order to accurately recover the broadband vibration signals from the strong interference of breathing and heartbeat. This project is a new research direction for non-contact vital sign detection. The contributions will provide valuable experience and technical support for the development of diagnosis technology in the field of medical and healthcare.