High Quality Integration
- By Eddie Reynolds
- Oct 06, 2020
Whether you are walking through U.S. customs at
an airport or the lobby at a critical facility like
a data center, a visitor checkpoint to verify your
identity is standard. In fact, visitor management
is an essential facet of the overall security solution, keeping
employees, guests and assets safe from threats.
While there are several choices for visitor management
platforms, biometric technologies have taken off over the last few
years, most specifically facial recognition which is expected to be
a $7 billion market by 2024. When compared to fingerprint or
retina scanners, facial recognition is considered a less intrusive
and more reliable way of tracking foot traffic for facilities for
Face detection allows enterprises to adopt a more seamless
and secure approach to visitor management, when compared
to solutions that solely rely on access cards and codes—which
are more easily manipulated. Whether an end user’s goal is to
recognize, identify, or verify a person, visitor management
solutions—such as those that use video surveillance systems
equipped with facial recognition software and integrate with
access control—can streamline visitor entry and allow enterprises
to create a “virtual perimeter.”
However, in order for these solutions to be effective, the
underlying video surveillance system must be set up to capture
video optimally. And if your surveillance deployment isn’t paired
with high-quality illumination, it will be difficult to guarantee
actionable analytics results.
Put simply, facial recognition video systems work by making
use of security cameras to scan images, using algorithms to map
faces and detect the features that make them unique. The system
then translates this information into hundreds of data points,
representing the geometry of one’s face.
These digital “face prints” are then used as reference points
when employees or visitors arrive at a facility, comparing those
who enter against a vast library. If the person matches the record
of an approved employee or visitor, they will be granted access to
the facility. Similarly, organizations can also use facial recognition
to blacklist individuals, barring them entrance or keeping track
of their presence and movements. However, the insights derived
from facial recognition analytics are only as good as the images
the system analyzes. Without clear, high-definition video, these
solutions can often fall flat.
Studies, some dating back to the late 1990s, have shown time
and time again that lighting plays a crucial role in the accuracy of
facial recognition systems. In fact, it has been argued that changes
in lighting conditions can make two images of the same person
seem less similar than two images of different people, according
to the National Institute of Standards and Technology in their
research and findings outlined in the report, Quantifying How
Lighting and Focus Affect Face Recognition Performance.
Variances in brightness and direction of lighting in reference
images can seriously hinder the detection capabilities and
accuracy of facial recognition.
DEPLOYING A SYSTEM
For example, a large corporate campus might be deploying
a video system with facial recognition software, the reference
images taken of each employee or visitor will most likely be
done in a well-lit space. While this yields a clear and accurate
reference point, these same conditions may not always be
present when the solution is in use. If the surveillance cameras
are placed outdoors at points of entry around the campus,
the availability, quality, and direction of the light will be everchanging.
This makes it difficult to guarantee the success of the
facial recognition results, thus preventing the video solution
from working correctly.
One of the most common challenges for surveillance cameras
is capturing usable footage in low or no light scenarios. Without
consistent, adequate illumination, even IP and Internet-of-Things
(IoT) cameras cannot effectively record clear enough images for
facial recognition to identify entrants or possible threats.
While some cameras come conveniently equipped with builtin
LEDs that encircle the lens, they often come with drawbacks.
The range for visible LEDs built into a camera is around 150 feet,
typically covering a 30-degree field of view (FOV) even though
a standard camera FOV is often 90-degrees. This creates “hot
spots” in the middle of the camera’s view and can cause a total
“white-out” of the rest of the image.
BEST PRACTICES AND KEY TAKEAWAYS
There is no “one-size-fits-all” lighting option for facial
recognition deployments, there are a few things end users and
system integrators should keep in mind. Consider these key
takeaways when deploying lighting for your visitor management
Angle of illumination. Consider is the angle of illumination
when deploying an external lighting solution. Every camera has a
unique FOV, making it important to choose a lighting option that
best matches the camera’s requirements.
External versus built-in. While convenient, cameras with builtin
LEDs are prone to hot spots, attract bugs, and are susceptible
to heat buildup. External illuminators, however, can minimize
heat accumulation and allow integrators to adjust the angle of
illumination and pair any given camera lens with the perfect
range/wavelength for the application.
White light versus IR. For deployments where full-color video
is critical day or night, like facial recognition or other analyticsheavy
applications, white light is the optimal choice.
This article originally appeared in the October 2020 issue of Security Today.