Use of Color Channels to Extract Heart Beat Rate Remotely from Videos

Abstract

Since last decades, photoplythesmography (PPG) signals that are extracted from the optical absorption in the tissues are increasingly being used for health diagnosis. Despite a good literature, there are different claims about the use of color channels as red, green and blue for extraction of PPG signal, i.e., pulse rate from the videos captured through high resolution cameras. In this article, we present the technique for extracting the heart beat rate (pulse rate) from the videos captured through the mobile cameras for all three color channels and thier analysis. Experiments were performed on a DMIMS database comprising 720 videos, out of which we used 25 videos for the analysis. The pulse rate estimated from the blue channel, was in good agreement with reference data extracted using an MP20 monitor, used as the gold standard. The findings of the present study demonstrated the non-invasive color intensity method for detection of pulse rate from the pre-recorded video of 30 seconds. The algorithm is tested on the DMIMS dataset which we have captured in uncontrolled setting. The green channel is proven to be statistically significant for the video recorded followed by red and then blue channel. The accuracy of the pulse extracted is still low because of low signal to noise ratio.We therefore conclude that the presented technique is best for pulse rate extraction through a blue channel followed by red and green channels respectively.