Introduction Accuracy of camera-based heart rate (HRcb) measurement is often impaired by artifacts, e.g. from motion. Distorted signals lead to erroneous HRcb and reduced confidence in remote photoplethysmographic measurements.
Methods To avoid erroneous HRcb, we investigated six signal quality indexes (SQIs) from the literature in terms of their effect size and improved conventional SQI usage by developing a technique to combine them for identifying and sorting out distorted signals. We call this technique SQI-filtering. The influence of SQI-filtering on HRcb accuracy (ACC) was investigated using the “Binghamton-Pittsburgh-RPI Multimodal Spontaneous Emotion Database” (BP4D+). All analyses were performed in three important color channels.
Results The most powerful SQIs were snrSQI, aCCmSQI and rdspSQI. The SQI-filter increased accuracies of all color channels. Largest improvements (up to +60%) were achieved in the green channel (RGB-G) resulting in 80% accuracy. The overall highest accuracy of 84% was reached in the hue channel (HSV-H). Motion-rich videos from BP4D+ benefited most from the developed SQI-filter.
Conclusion The presented methodology helps to identify and discard distorted signals. This enables more reliable HRcb data in further applications and increases confidence in remote photoplethysmographic measurements.