Remote Detection of Heart Beat and Heart Rate from Video Sequences

Abstract

The article presents a proposal to detect the heart beat and an estimation of the instantaneous heart rate based on a video, using the variation of skin tone as a function of the blood flow, which are imperceptible to the human eye. This method of data acquisition does not require a sensor to be in contact with the user’s skin, so it is a non-invasive method, easy to acquire and can be performed from a certain distance. Using computer vision, it was possible to track the face over time using the KTL classifier to increase tracking accuracy and decrease the processing time. With the detection performed over time it was possible to visualize the variation of skin tone as a function of blood flow through a temporal filter and an equalization of the histogram. The final result was obtained by evaluating the histogram resulting from each processed frame of the video.