ESTIMATING CARDIAC PULSE RECOVERY FROM MULTI-CHANNEL SOURCE DATA VIA CONSTRAINED SOURCE SEPARATION

What is disclosed is a system and method for recovering a patient's cardiac pulse rate from a sequence of video images recording of that patient. In one embodiment, a reference signal of a particular frequency is generated at predetermined frequency intervals and a constrained source separation is performed on the source data to obtain an estimation of the source signal intended to be recovered. The reference signal is updated and constrained source separation is again performed. These operations are repeated for all frequencies of the reference signal. The frequency at which a minimum error is achieved is determined to be the subject's recovered cardiac pulse frequency. In such a manner, the source signal is extracted and recovered reliably from captured multi-channel RGB signals or multispectral signals. The teachings hereof find their uses in a variety of medical solutions including various military, security and telemedicine applications. Various embodiments are disclosed.