Stress classification using multispectral imaging technology

The stress classification algorithms are proposed based on multispectral imaging (MSI) system in this study. By extracting spectral features from the facial region, it is feasible to arrive at some quantity which can be related to the physiological property or functionality. The proposed method has been tested, and the accuracy for the classification of physical stress (PS) and emotional stress (ES) is over 90% by using the MSI techniques. Therefore, the classification feature of emotional stress and physical are successfully obtained. Experiments show that the classification result is encouraging, and the highly successful rate stems from the fact that the manipulated spectral quantity agrees with the physiological ground truth very closely. This study can lead to a practical system for the noninvasive classification of stress states.