The non-contact method is used to detect the physical stress of the human body, thereby providing a good model for practical application. However, the existing method is highly inaccurate and requires people’s background information. This study introduces a new method to detect the physical stress of the human body by maximizing thermal infrared imaging. First, the method of multi-object correlation is used to determine the region of interest that the face is sensitive to physical stress. Furthermore, the stress signal is extracted through the thermal signal and the method based on blind source separation is adopted to convert the thermal signal into an independent component that is sensitive to stress. Second, the independent component signal is amplified to extract the peak frequency. Lastly, the baseline and physical stress states are classified and identified using the deep learning algorithm model. The algorithm model was verified with respect to the physical stress ground truth to classify the baseline and physical stress status. The algorithm achieved improved results in the experiment with an accuracy rate of 90%, thereby providing a foundation for future industrialization. Experimental results demonstrated that thermal imaging, as a non-invasive method without background information, has the potential to detect physical stress among humans. This demonstration is the first thermal imaging-based method for contact-free physical stress detection.

中文翻译：

---

使用热成像和盲源分离的非接触式物理应力测量

非接触式方法用于检测人体的物理压力，从而为实际应用提供了良好的模型。但是，现有方法非常不准确，并且需要人们的背景信息。本研究介绍了一种通过最大化红外热成像来检测人体物理压力的新方法。首先，使用多对象相关方法确定面部对物理压力敏感的感兴趣区域。此外，通过热信号提取应力信号，并采用基于盲源分离的方法将热信号转换为对应力敏感的独立分量。其次，独立分量信号被放大以提取峰值频率。最后，使用深度学习算法模型对基线和身体压力状态进行分类和识别。相对于物理应力地面真实性验证了算法模型，以对基线和物理应力状态进行分类。该算法在实验中获得了改进的结果，准确率达到90％，从而为以后的工业化奠定了基础。实验结果表明，热成像作为一种无背景信息的无创方法，具有检测人类身体压力的潜力。该演示是第一个基于热成像的无接触物理应力检测方法。相对于物理应力地面真实性验证了算法模型，以对基线和物理应力状态进行分类。该算法在实验中获得了改进的结果，准确率达到90％，从而为以后的工业化奠定了基础。实验结果表明，热成像作为一种无背景信息的无创方法，具有检测人类身体压力的潜力。该演示是第一个基于热成像的无接触物理应力检测方法。相对于物理应力地面真实性验证了算法模型，以对基线和物理应力状态进行分类。该算法在实验中获得了改进的结果，准确率达到90％，从而为以后的工业化奠定了基础。实验结果表明，热成像作为一种无背景信息的无创方法，具有检测人类身体压力的潜力。该演示是第一个基于热成像的无接触物理应力检测方法。具有检测人类身体压力的潜力。该演示是第一个基于热成像的无接触物理应力检测方法。具有检测人类身体压力的潜力。该演示是第一个基于热成像的无接触物理应力检测方法。