Local signal-to-noise peak detection
Biometric sensor systems are provided for identifying fundamental heart rate harmonics within noisy sensor signals. The system calculates a local signal-to-noise ratio for one or more identified frequency bands received in a biometric signal. The identified frequency bands are ranked based upon the calculated local signal-to-noise ratio. The fundamental heart rate is identified based upon the ranking of the identified frequency bands.
1. A computer system for identifying fundamental heart rate harmonics within noisy sensor signals, comprising:
one or more processors; and
one or more computer-readable media having stored thereon executable instructions that when executed by the one or more processors configure the computer system to perform at least the following:
receive, at a sensor input interface, biometric signal data from a sensor in contact with a user;
extract, using a computer processor, a frequency-domain biometric signal from the biometric signal data;
identify a candidate frequency band within the frequency-domain biometric signal;
identify one or more harmonics associated with a candidate peak, wherein the one or more harmonics and the candidate peak are within the candidate frequency band;
generate a summed local signal value by summing an amplitude of the candidate peak with amplitudes of each of the one or more harmonics;
generate a summed local noise value by summing amplitudes associated with signals that occur between the one or more harmonics and dividing the summation of amplitudes by a number of bins utilized in the summation of amplitudes;
calculate a candidate figure of merit by dividing the summed local signal value by the summed local noise value;
identify a fundamental harmonic within the frequency-domain biometric signal based upon the candidate figure of merit; and
display a user's heart rate based upon the identified fundamental harmonic.
2. The computer system of claim 1 , wherein the biometric signal data comprises a photoplethysmogram.
3. The computer system of claim 1 , wherein the biometric sensor data comprises a photodetector.
4. The computer system of claim 1 , wherein the executable instructions include instructions that are executable to configure the computer system to:
identify another frequency band within the frequency-domain biometric signal;
identify one or more other harmonics associated with the other frequency band;
generate another summed local signal value by summing an amplitude of the other frequency band with amplitudes of each of the one or more other harmonics;
generate another summed local noise value by summing amplitudes associated with signals that occur between the one or more other harmonics;
calculate another candidate figure of merit by dividing the other summed local signal value by the other summed local noise value;
rank the candidate peak and the other peak based upon their respective figure of merit values; and
identify the fundamental harmonic within the frequency-domain biometric signal by identifying a frequency band associated with the highest ranked figure of merit.
5. A method for identifying fundamental heart rate harmonics within a noisy photoplethysmogram, the method comprising:
receiving, at a sensor input interface, a noisy photoplethysmogram from a photodetector in contact with a user;
extracting, using a Fast-Fourier Transform, a frequency-domain biometric signal from the noisy photoplethysmogram;
identifying a first candidate frequency band within the frequency-domain biometric signal;
calculating a first candidate figure of merit by dividing a first summed local signal value by a first summed local noise value, wherein:
the first summed local signal value comprises a summation of an amplitude of a first candidate peak with amplitudes of each of one or more first harmonics of the first candidate frequency band, and
a first summed local noise value comprises a summation of amplitudes associated with signals that occur between the one or more first harmonics divided a number of bins utilized in the summation of amplitudes;
identifying a second candidate frequency band within the frequency-domain biometric signal;
calculating a second candidate figure of merit by dividing a second summed local signal value by a second summed local noise value, wherein:
the second summed local signal value comprises a summation of an amplitude of a second candidate peak within the second candidate frequency band with amplitudes of each of one or more second harmonics of the second candidate frequency band, and
a second summed local noise value comprises a summation of amplitudes associated with signals that occur between the one or more second harmonics;
ranking the first candidate frequency band and second candidate frequency band based upon the first candidate figure of merit and the second candidate figure of merit;
identifying a fundamental harmonic heartbeat associated with the user within the noisy photoplethysmogram based upon the ranking; and
displaying the user's heart rate based upon the fundamental harmonic heartbeat associated with the user.
6. The method as recited in claim 5 , wherein extracting the frequency-domain biometric signal from the noisy photoplethysmogram comprises generating the frequency-domain biometric signal within a plurality of bins.