Noncontact pulse wave detection by two-band infrared video-based measurement on face without visible lighting

In this paper, we propose a novel noncontact pulse wave monitoring method that is robust to fluctuations in illumination through use of two-band infrared video signals. Because the proposed method uses infrared light for illumination, the method can be used to detect a pulse wave on a human face wit...

Full description

Saved in:
Bibliographic Details
Published inArtificial life and robotics Vol. 23; no. 3; pp. 345 - 352
Main Authors Mitsuhashi, Ryota, Okada, Genki, Kurita, Koki, Kagawa, Keiichiro, Kawahito, Shoji, Koopipat, Chawan, Tsumura, Norimichi
Format Journal Article
LanguageEnglish
Published Tokyo Springer Japan 01.09.2018
Springer Nature B.V
Subjects
Adaptive filters
Artificial Intelligence
Bandpass filters
Computation by Abstract Devices
Computer Science
Control
Error detection
Ground truth
Hemoglobin
Illumination
Infrared radiation
Light
Lighting
Mechatronics
Original Article
Pixels
Pulse rate
Robotics
Shading
Signal processing
Variations
Video signals
Pulse rate
Pulse wave
Noncontact measurement
Infrared
Online AccessGet full text

Cover

More Information
Summary:In this paper, we propose a novel noncontact pulse wave monitoring method that is robust to fluctuations in illumination through use of two-band infrared video signals. Because the proposed method uses infrared light for illumination, the method can be used to detect a pulse wave on a human face without visible lighting. The corresponding two-band pixel values in the video signals can be separated into hemoglobin and shading components by application of a separation matrix in logarithmic space for the two pixel values. Because the shading component has been separated, the extracted hemoglobin component is then robust to fluctuations in the illumination. The pixel values in the region of interest were spatially averaged over all the pixels of each frame. These averaged values were then used to form the raw trace signal. Finally, the pulse wave and the corresponding pulse rate were obtained from the raw trace signal through several signal processing stages, including detrending, use of an adaptive bandpass filter, and peak detection. We evaluated the absolute error rate for the pulse rate between the estimated value and the ground truth obtained using an electrocardiogram. In the experiments, we found that the performance of the proposed method was greatly improved compared with that of conventional methods using single-band infrared video.
ISSN:1433-5298
1614-7456
DOI:10.1007/s10015-018-0430-5