I first was exposed to biometrics at scale when I was working within the National Strategy for Trusted Identities in Cyberspace – Identity Ecosystem Steering Group that would host meetings “next to” biometric industry conferences. They were really kinda freaky for the technology presented and how I imagined it all being used.
I still have a very cautionary attitude towards biometrics technology however I am not scared of them in part because I have met folks who work more closely with and deepened my understanding.
In the fall of 2020 I proposed to Jack Callahan. We put on the first Thoughtful Biometrics workshop in March 2021. Now we are organizing the 2nd one coming up March 16, 2023. It is virtual and online. Learn more here and Register!
This is explanation of Biometrics I wrote for the workshop.
The term “biometrics” comes from two roots:
bio meaning biological life + metrics meaning measurement
Therefore, it is defined as the measurement of biological characteristics. Every person has unique bodily or physical (biological) characteristics, and these can be measured, described and recorded or documented in various ways. Once documented they can be compared for identification or authentication purposes.
Biometrics existed well before the creation of digital computers. For example, fingerprints have been collected and compared to establish identity since the mid-1800’s, and photographs used on documents such as passports starting about 100 years ago.
Digital technology and other key innovations over the last 40 years, including advances in computer vision, algorithmic processes, and matching algorithms have expanded biometrics deeply into our digital world.
Types of biometrics
All biometric modalities are basically two types; physiological and behavioral.
Examples of physiological biometrics include fingerprint, iris, retina, face, palm, and vein recognition. Examples of behavioral biometrics include signature and voice recognition.
APPLICATIONS OF BIOMETRICS:
1:1 matching for authentication:
Looking at a biometric associated with a particular person or record and seeing if the person presenting a sample of their biometric matches the biometric template that was previously enrolled. For example, 1:1 Matching on devices such as unlocking a phone using a fingerprint or faceprint.
Human 1:1 Matching against a document such as a passport, drivers license or other document. This can be done by a human looking at the picture and comparing, such as when you buy an age-restricted product. The clerk does a match between the document and a face along with checking the age of the person presenting the document.
Computer 1:1 Matching against a document such as a passport, drivers license or other document. Computer vision is used to look at the person standing in front of a camera and compare them to a photo encoded on a document.
1:1 Matching against a sample enrolled in a system. An employer might have a biometric enrolled and on file to be compared to 1:1 when the employee presents themselves.
1:small n matching for authentication
This type of biometric usage involves the enrollment of a small group of people into a system. One use-case is all the people boarding a particular plane. The individuals traveling can share their photo and travel documents with the airline in the check in process. The photos are now in a gallery of a few hundred. When going to board the plane individuals can present themselves and be allowed to board after they match one of the faces in the gallery of photos of people who are passengers on the plane.
1:large n matching and identification
An important question for this mode is: does it happen in real time, or later in a forensic context?
In real time a video can capture snapshots of faces and then run the faces against a large n of potentially millions of people. If implemented on a camera on a public street, this would output a list of all the people who walked past.
Forensic use of biometrics really began over a century ago with fingerprint matching. Fingerprints lifted from crime scenes were compared with fingerprint records of people. This is still done today but with electronic systems doing the matching. For forensic facial recognition, images of people are captured in retrospect from video or still photographs from a crime scene. These are compared with large data sets of images of people along with their names.
EXAMPLES OF OTHER PLACES WE USE BIOMETRICS
Smart Speakers
Voice recognition goes beyond understanding speech; some devices are able to distinguish between people based on a pre-enrolled voice print. We could also see databases of voice prints created to compare samples against, much like we have photos of people being compared using facial recognition algorithms.
Wearable Sensors for Health tracking
There are more and more devices that use sensors to track things like heart rate, temperature, or oxygen levels. These devices in the form of rings, watches, bracelets, etc. record and share this information.
Augmented Reality (AR) and Virtual Reality (VR) Devices
AR and VR work with sensors that track head location and movements, eye movement and focus, facial expressions, hand movements, heart rate and even perspiration.
We may have missed some biometrics use-cases – feel free to reach out and share more!