Emotional State Estimation from the Non-Contact, Automatic and Webcam-Based Human Cardiac Pulse

摘要

Inferring emotional states from internal physiological signals has received considerable attention as these signals could directly and objectively reflect emotional states. However, precise measurement of the physiological signals (such as human cardiac pulse) requires people to wear some chest straps or adhesive gel patches. It is therefore inconvenient to measure signals in the daily life. The elaborated technique presented here has achieved efficient pulse rate estimation based on the face image captured by the non-contact, automatic and webcam-based method. The desired image signal was obtained by independent component analysis and proper channel selection. Then after local mean decomposition and feature selection, a set of features are obtained for emotional state estimation using support vector machine. The experiment result shows that it is possible to estimate human's emotional states based on the technique presented here. The proposed technique may be useful for monitoring person for the purpose of health care, psychological testing or security.