基于生理微动的废墟压埋人体高分辨生物雷达成像技术研究

Bioradar is widely used in rescue of human buried under rubble after disaster because of its penetrability and non-contact detection. Since the detection objects of present bioradars are vital signs (respiration rate and heart rate), high detection false alarm rate of human subjects under rubble is inevitable. If high-resolution imaging of human could be realized, the above problems can be expected to be solved. However, the defocusing problem caused by rubble and the low spatial resolution caused by low center frequency are the bottleneck problems of this research. In this project, low-frequency ultra-wideband (UWB) multiple input multiple output (MIMO) bioradar is our basic technology, and the specification of the rubble structure and human under rubble is selected as our starting point. Firstly, rubble structure model suitable for survival of human under rubble is built. Then, a priori physiological characteristics of the human under rubble are analyzed and utilized. Next, the autofocusing method using the above prior information is utilized to form focused image. Finally, the micro-doppler characteristics are extracted and quantized into enhancement factor and utilized to form resolution-enhanced image of human under rubble. Through the above efforts, this project will provide new theories and techniques for high-resolution imaging of human under rubble and promote the further development of the bioradar technology.

生物雷达具有可穿透障碍物、非接触探测等优点，在灾后废墟压埋人员搜救中具有重要作用。由于现行生物雷达以生命参数（呼吸率、心率等）作为探测指标，压埋人体探测的虚警率偏高。如能实现压埋人体的高分辨成像，该问题可迎刃而解。废墟介质引起的成像散焦、低中心频率引起的成像分辨率下降是废墟压埋人体高分辨成像的技术瓶颈。本项目拟以低频超宽带（UWB）多输入多输出（MIMO）生物雷达技术为基础，从废墟结构和压埋人体的特殊性入手，首先建立适于压埋人体存活的废墟结构模型，并分析废墟压埋人体的生理微动特征，然后将上述先验信息结合到自聚焦技术中，实现废墟结构参数补偿和人体生理微动信号的聚焦成像，以此为基础在空间-时间-多普勒联合图像域提取压埋人体的微多普勒特征，并将其量化为增强因子，实现压埋人体的分辨率增强成像，为废墟压埋人体成像探测提供新的理论和技术方法，提高生物雷达技术的实用性。

地震、塌方等自然灾害严重威胁人类生命安全。如何在灾后搜救行动中准确、快速地发现废墟压埋幸存者并挽救其生命，是全球普遍关注的问题。生物雷达技术是穿透介质探测人体生命体征信号的重要手段之一。项目主要针对多输入多输出（MIMO）生物雷达生命探测成像中电磁波穿透介质所引起的图像散焦和成像分辨率下降等问题展开研究。建立了人体生命体征信号的回波模型和MIMO成像模型，对典型介质穿透场景下的生命体征信号回波进行了电磁仿真，分析了介质对成像的影响机理，研究了人体生命体征信号的介质穿透补偿自聚焦成像技术、基于深度学习的介质穿透补偿成像技术，提出了微弱生命体征信号的多增强方法融合框架及人体目标高维增强成像方法、穿透条件下人体与动物识别方法、多人体目标自动跟踪方法等系列方法。研发了UHF波段、S波段MIMO生物雷达系统，并开展了一系列的穿透探测成像实验，涵盖了自由空间、穿透单层介质、穿透多层介质、穿透模拟废墟等多个场景，实现了穿透条件下人体目标的高维成像，验证了方法的可行性，显著提高了穿透条件下的成像质量，为高性能生命探测装备的研发提供新的理论方法支撑和技术途径，有助于提高生命探测雷达的实战化水平，为挽救更多的生命做出贡献。