Below is information to help you understand more about the technology and hopefully answer questions you might have regarding biometrics.
In the context of the information below we generally will be referring to fingerprint based biometric systems and at times specifically
referring to the systems utilized by Human Element Biometrics.
What is Biometrics?
Biometrics is the science of recognizing the identity of a person passed on the physical or behavioral attributes of the individual such as fingerprints,
face, voice, iris, etc. Biometric systems operate under the premise that many of the physical or behavioral characteristics of humans:
- Are distinctive to an individual
- Can be reliably acquired via appropriately designed sensors or equipment
- Can be represented in a format that lends itself to automatic decision-making in the context of identity management
A Simple Solution
It sounds really complicated, and no doubt the principles behind the underlying technology are very detailed and quite scientific. But the application
of biometrics is quite simple – You put your finger on a fingerprint reader and your identity is verified.
Benefits of Applying Biometrics
• Security
• Loss/Fraud Prevention
|
• Accountability
• Customer Confidence
|
• Convenience
• Employee Confidence
|
• Cost Savings
|
How it Works
Our fingerprint solutions work in two steps – Enrollment (or fingerprint registration) and Verification (or fingerprint matching).
Enrollment/Registration: The enrollment process consists of creating an enrollment/registration fingerprint template. A template is the features and
characteristics of your fingerprint that has been processed via a fingerprint feature algorithm. The process of creating an enrollment template for our
applications consists of placing your finger on the fingerprint sensor four separate times creating four samples. The four samples are then compared and
merged to form a single consistent sample and the sample is then processed against the algorithm to create the registration template. Once the registration
template is created it is persisted (stored) in a database, file system, etc, indexed by a unique identifier that represents that person (for example; an
account number, social security number, employee number, etc.)
Verification/Matching: To perform fingerprint verification, the enrollment/registration template for the person being verified (created and stored in
the enrollment process) will first be retrieved from the database, file system, etc. The person being verified will then place their finger on the fingerprint
sensor one time to create a sample. The sample is then processed against the algorithm to create the verification template. The verification template is
compared to the registration template to determine if enough of the same features exist between the two samples to declare a match.
|
Reliability and Accuracy
Our systems utilize the fingerprint readers and recognition algorithm developed by DigitalPersona. The algorithm incorporates traditional fingerprint identification methodologies, creating each user’s unique identifying information for recognition. With over ten years of study, extensive research, and testing, DigitalPersona’s recognition engine is one of the most robust fingerprint recognition algorithms available today.
The performance of fingerprint algorithms is measured primarily as a tradeoff between two attributes:
False Acceptance Rate (FAR) which is the probability that an intruder will be accepted by the system.
False Rejection Rate (FRR) which is the probability that a legitimate registered fingerprint user will be incorrectly rejected by the system.
The FAR and FRR rates and the accuracy of the system are a direct result of the quality of the fingerprint of the individual user. Testing with large groups of people over an extended period has shown that a majority of all users have such feature-rich fingerprints that they will virtually always be recognized accurately by our engine and practically never obtain a false acceptance or a false rejection. Our recognition engine is optimized to recognize prints of poor quality. However, a very small number of fingerprints are either worn from manual labor or have unreadable ridge lines and are very difficult to match. A small fraction of users may sometimes have to try a second or even a third time to obtain an accurate reading.
A major advantage of the DigitalPersona algorithm over those of its competitors is that it employs an enhanced version of the raw image that comes from the fingerprint reader. In addition to making better use of the fingerprint image provided by the fingerprint reader, the DigitalPersona algorithm benefits from proprietary image processing, pattern recognition, and statistical techniques. This improves the results obtained from poor-quality dry, damaged, and minutia-impoverished prints, and blurred or skewed print images.
The key benefit of our system is that it brings together ease of use and reliability. The system is entirely rotation-invariant, meaning that the user can put their fingerprint onto the fingerprint reader at any angle. It also provides extremely low FAR and FRR.
Privacy
Because the information regarding the fingerprint is only a template of its features and characteristics, privacy is preserved. Only the template is stored. The fingerprint image is never stored or retained. Just as important, the template cannot
be reverse engineered to reproduce the fingerprint image.
For more information see the section:
Privacy and Concerns
Spoofing
The fingerprint readers and underlying software we deploy has an anti-spoof technology that can detect and reject the most likely attempts to fool it. That is, the reader is highly effective at rejecting photocopies of fingerprints, photographs of fingerprints, and attempts to re-submit any latent fingerprint image on the reader window through dusting, flashlights, misting, etc.
There are Internet reports which describe complex procedures for fooling fingerprint readers in general. These procedures are complex and by no means simple, quick or easy. These methods almost always involve lifting a print, enhancing it, scanning it, digitally processing the image, printing a film of the image, etching a printed circuit board with strong acids, then molding a 3-D model of the finger with gelatin or other substances. Such processes are hardly practical for a malicious hacker. In almost all reported cases the person whose fingerprint was spoofed cooperated in the process. Nevertheless, as with any security technology, with enough effort and financing, a carefully crafted 3-D model of a finger will sometimes be accepted by every fingerprint reader on the market today. It is important to note that this susceptibility applies to all fingerprint reader technologies: optical and semiconductor, be it swipe-style or placement readers.