An image measurement heart rate application in video telephony systems

Video telephone systems are convenient media for people to communicate remotely in life. If the video can be analyzed in the process of video chat to extract the relevant physiological indicators of the communicators such as emotions or stress, it can help the communicators to make proper conversations. In this paper, we propose (1) Similar Brightness Area Selection Method (2) and Difference-driven Weight-Setting Method to measure the heart rate in the non-contact image mode that can be adapted to the video signal in the videophone. After the image is compressed, it is transmitted through the network. After being transmitted to the video receiver for image decompression, we also propose that the luminance (Y) instead of Green channel (G) of the image should be utilized based on Photoplethysmography (PPG) to infer the heart rate. In our experiment, we have adjusted the weight range of our Difference-driven Weight-Setting Method for four commonly used video compression codecs including H.264, H.265, VP8, and VP9 together with five different compression bit rates and the results have shown to be effective in improving the accuracy of the heart rate measurement. The root mean square error (RMSE) of the heart rate measurement has also been calculated as evaluation metrics in the experiment. The mean and standard deviation of the overall RMSE is 2.14±0.52 bpm (beats per minute) that have shown the feasibility of our proposed methods to effectively and accurately measure the heart rate in the video telephony system.