Remote Heart Rate Determination in RGB Data - An Investigation using Independent Component Analysis and Adaptive Filtering

Abstract

An emerging topic in the field of elderly care is the determination and tracking of vital parameters, such as the heart rate. This parameter provides important information about a person's current health status. Within the last years, various research focussed on this topic. The recognition of vital parameters is increasingly relevant for our ageing society. This paper presents a method to remotely determine the human heart rate with a camera. At this point, we suggest to use independent component analysis (ICA) and adaptive filtering for a robust detection. In our processing chain, we used different image processing techniques, e. g. face detection, and signal processing techniques, e. g. FFT and bandpass filtering, in this study. An evaluation with several probands, illuminations, frame rates and different heart rate levels showed that we could achieve a mean error of 4.36 BPM, which corresponds to CAND of 94.45 %, and a speed of 35 fps.

Real-Time Heart Rate Estimation using Fast Independent Component Analysis and Hangover-Time Filter

“…For eliminating false detections appearing due to sudden changes in the heart rate or sudden changes in data intake from one time to the next time, [1] suggested in using an adaptive filter. The working principle of the adaptive filter is to check the absolute difference between the current heart rate value ( ) 0 HR t and the previous heart rate value ( )…”

“…The test ranged the average heart rate from 55 BPM to 120 BPM. The mean error is calculated by using the equation provided in [1]. Equation (12) calculates the mean error for each time from…”

“…Implementing a new PPG-based method of BSS called FastICA was developed by Wiede et al [1]. FastICA reduces complex multiplication when compared to traditional ICA.…”

“…Using the framework proposed in [1], this paper implements a real-time fast ICA contact-less PPG method using OpenCV library in Python environment for detecting a person's heartbeat. Moreover, this paper proposes a hangover-time filter to further improve system performance against motion-induced artifacts.…”

“…This paper also proposes on using a Hangover-Time filter for robust detection. Following the recent research [1], this project also utilizes OpenCV for face detection and isolating the forehead region, Fast Fourier Transform Analysis and Band-Pass Filter. Compared to the adaptive filter, the hangover-time filter has a difference in a mean error of ~0.26BPM and the corresponding Complement of the Absolute Normalized Difference (CAND) of 95.05% is achieved.…”