Instrumentation & Measurement Magazine 25-1 - 20

Contactless Remote Assessment
of Heart Rate and Respiration Rate
Using Video Magnification
Bruce Wallace, Leen Yassin Kassab, Andrew Law, Rafik Goubran, and Frank Knoefel
R
esearch into methods to assess heart rate and respiration
rate through non-contact means has been
ongoing for many years [1], and this work has expanded
in scope in recent years following the publication of
two methods in 2012 and 2013, respectively: video magnification
(VM) [2] and remote photoplethysmography (rPPG) [3].
Both methods are founded on similar core principles of exploiting
subtle changes within a video frame in either color,
motion or a combination of the two, followed by signal processing
to enhance and magnify these changes. This has
allowed for non-contact methods to detect skin color changes
within the pulse wave for heart rate measurement or motion of
the chest associated with respiration. This work reviews recent
advances and relevant works associated with the application
of these techniques for non-contact sensing that supports remote
assessment of heart rate and respiration rate. Remote
assessment requires an effective means for non-contact assessment
that also allows the separation of the sensing from
the assessment processing and/or a user of the assessment
through supporting communications networks so that each
can be potentially at different locations.
Many researchers have worked to build on the proposed
VM [2] or rPPG [3] methods independently, and only recently
have these two paths of innovation been brought together. Although
initially distinct, these two models have evolved into a
family of methods that now have many attributes and aspects
in common, and the distinction between any two implementations
of non-contact heart/respiratory rate measurement is
based on the choices made within each algorithmic step. As
such, methods presented as an implementation of VM may
have significant similarities to one that is based on rPPG, while
two implementations of VM, or rPPG, may be significantly
different.
Vital Sign Assessment
Vital sign measurements, including heart rate, respiration
rate and body temperature, are often used for health assessments
by clinicians and require direct contact with a patient
in a clinical environment. Measurement of vital signs through
20
non-contact and, more specifically, remote means has many
potential applications. In the public health domain, the risk of
spread of infectious disease can be higher at locations where
people gather (e.g., transportation terminals, stadiums, etc.)
or in locations where older and frailer individuals are congregated,
such as long-term care facilities or communal senior
living residences. Contactless remote vital signs measurement
could also improve health assessment for remote populations
that live long distances from care centers, or as the recent
COVID-19 social isolation precautions have demonstrated, for
the delivery of day-to-day care to patients without requiring
in-person assessment at a clinic or hospital.
Non-contact thermal assessment of body temperature has
been widely adopted in response to the COVID-19 pandemic
[4]. Although this method is effective for rapid fever detection,
it is limited to the assessment of that single vital sign. By adding
non-contact assessment of heart rate and respiration rate, a
more holistic assessment of health would be possible. Methods
for remote assessment must be minimally intrusive in terms of
the privacy of the individual and the flow of people in public
spaces, while also automating the analysis to minimize direct
interaction/engagement with trained personnel.
The VM [2] and rPPG [3] algorithms were each proposed
as methods for non-contact, remote vital sign assessment
through analysis of video, and they share a common processing
model that is summarized in Fig. 1. The first step is the
capture of video of the subject and may involve either realtime
processing of live video or offline processing of video
that was previously captured. VM [2] proposed the use of the
Laplacian pyramid for spatial decomposition of the region of
interest (ROI) followed by temporal processing of the decomposed
images. rPPG [3] focused on the specular reflection of a
known light source and proposed the use of two orthogonal
color channels formed from two combinations of the RGB signals.
The original papers proposed the use of full color (RGB)
video, and recent works have expanded the methods to operate
on video from monochrome, thermal and InfraRed (IR)
cameras [5]-[7]. Once captured, the video first undergoes spatial
processing to determine, analyze and track an ROI within
IEEE Instrumentation & Measurement Magazine
1094-6969/22/$25.00©2022IEEE
February 2022
Instrumentation & Measurement Magazine 25-1

Table of Contents for the Digital Edition of Instrumentation & Measurement Magazine 25-1
