The invention relates to a non-contact wink and
heart rate combined detection
system and method based on second-order
blindness identification. When subjects are at a natural relaxed state, face video data comprising eyes is acquired, after original video data is acquired, a detection face area is selected, the video data is preprocessed, six-channel signals after preprocessing are obtained, and the signals are marked as x=[xR1, xR2, xG1, xG2, xB1, xB2]T; target areas in continuous frames are obtained by use of a Meanshift
algorithm, and the six-channel signals are obtained by use of the same method; by taking original signals after the preprocessing as input signals, blind
source separation is carried out by use of a second-order blind identification
algorithm, source signals after the separation are marked as y=[y1, y2, y3, y4, y5, y6]T;
signal identification and screening are performed on the separation signals y obtained in the step S101, and needed wink and BVP signals are selected by use of a kurtosis-based separation component automatic identification method; and a wink frequency and a time length of the obtained wink signals are calculated, power
spectrum analysis is performed on the obtained BVP signals, and a
heart rate estimation value is obtained. The method has the advantages of high accuracy, high anti-interference capability, high
algorithm efficiency and the like.