| 摘要 |
What is disclosed is a computationally efficient system and method for estimating a subject's cardiac pulse rate from multi-channel source video data. In one embodiment, A time-series signal is continuously processed by repeatedly: (1) conditioning the estimated source signal obtained on a previous iteration to produce a next reference signal; (2) using this reference signal to perform constrained source separation on this next segment to obtain an estimated source signal. A frequency at which this next estimated source signal converged is the subject's estimated cardiac pulse rate for this signal segment. The reference signal is repeatedly updated. Upon convergence, the sliding window is shifted to define a next segment of the time-series signal. The method repeats for each time-series signal segment on a continuous basis or until a termination criteria is met. In such a manner, the subject's cardiac pulse rate is estimated from video data on a continuous basis. |
