Final published version
Research output: Contribution in Book/Report/Proceedings - With ISBN/ISSN › Chapter (peer-reviewed) › peer-review
Research output: Contribution in Book/Report/Proceedings - With ISBN/ISSN › Chapter (peer-reviewed) › peer-review
}
TY - CHAP
T1 - Video-Based Heart Rate Detection: A Remote Healthcare Surveillance Tool for Smart Homecare
T2 - Advanced Sciences and Technologies for Security Applications
AU - Harrison, Thomas
AU - Zhang, Z.
AU - Jiang, R.
PY - 2022/9/9
Y1 - 2022/9/9
N2 - A novel approach to extract a heart rate signal from video footage consisting of a five stage processing pipeline is presented. Two extraction methods were used to obtain a heart rate. The first used the Fast Fourier transform to estimate an average heart rate by peak frequency analysis in the frequency distribution and estimated heart rates with a MAE as small as 2.32 BPM. This MAE value is smaller than those found by previous research which used PPG signals and BCG signals to extract a heart rate. The second approach used the Short-time Fourier transform to produce a time series of heart rate estimation which, when compared to accepted ground truths produced a covariance value of up to 0.9206335. Using a hybrid CNN-LSTM model an ECG-like signal was extracted from time-series heart beat waveforms. The resultant ECG-like signal displayed some of the characteristic ECG traits however it was not stable across the entire time period. Potentially, such a non-invasive heart monitoring can serve as a remote healthcare surveillance tool for smart homecare. © 2022, Springer Nature Switzerland AG.
AB - A novel approach to extract a heart rate signal from video footage consisting of a five stage processing pipeline is presented. Two extraction methods were used to obtain a heart rate. The first used the Fast Fourier transform to estimate an average heart rate by peak frequency analysis in the frequency distribution and estimated heart rates with a MAE as small as 2.32 BPM. This MAE value is smaller than those found by previous research which used PPG signals and BCG signals to extract a heart rate. The second approach used the Short-time Fourier transform to produce a time series of heart rate estimation which, when compared to accepted ground truths produced a covariance value of up to 0.9206335. Using a hybrid CNN-LSTM model an ECG-like signal was extracted from time-series heart beat waveforms. The resultant ECG-like signal displayed some of the characteristic ECG traits however it was not stable across the entire time period. Potentially, such a non-invasive heart monitoring can serve as a remote healthcare surveillance tool for smart homecare. © 2022, Springer Nature Switzerland AG.
KW - Artificial intelligence of things
KW - Heartrate monitoring
KW - Smart homecare
U2 - 10.1007/978-3-031-04424-3_10
DO - 10.1007/978-3-031-04424-3_10
M3 - Chapter (peer-reviewed)
SN - 9783031044236
T3 - Advanced Sciences and Technologies for Security Applications
SP - 159
EP - 195
BT - Big Data Privacy and Security in Smart Cities
A2 - Jiang, Richard
A2 - Bouridane, Ahmed
A2 - Li, Chang-Tsun
A2 - Crookes, Danny
A2 - Boussakta, Said
A2 - Hao, Feng
A2 - Edirisinghe, Eran A.
PB - Springer
CY - Cham
ER -