False acceptance refers to an unauthorized user given access to a system which he is not allowed to access. Therefore a false acceptance rate of 1% means the system will incorrectly allow access to someone who is not allowed 1% of the time. I have tried to google around but cannot come up with how the false acceptance rate is determined. Does it refer to a random error as explained in scenario 1 or to a systematic error as explained in scenario 2?

Scenario 1. Person A is authorised. Person B is not authorised. Person B has a very different fingerprint from person A. However, person B tries to authenticate 100 times. Out of those 100 times, 1 time is successful.

Scenario 2. Person A is authorised. Person C is not authorised. However, person C belongs to 1% of the population that have very similar fingerprint as person A. Therefore, person C will always trigger a false acceptance in the system.

Assumptions: Only 1 fingerprint, person A, is registered with the system. Therefore, it is impossible for person B,C to match with other prints.

Edit: In short, does the false acceptance rate refer to accuracy or precision?

  • 1
    So in short your asking how accurate biometrics are vs how precise they are, specifically regarding fingerprints - do you have a source for that 1%? - either way, fingerprints aren't an absolute black and white - it's if a reading is close enough to that saved on file - and this is a configurable parameter that depends on the implementation. it's a tradeoff between security and ease of use. Commented Dec 12, 2014 at 3:33
  • The 1% is an arbitrary number I made up to illustrate the point. Yep, its accuracy vs precision. Will edit the question to reflect that. Commented Dec 12, 2014 at 3:35
  • "I have tried to google around but cannot come up with how that 1% is determined." - yet you claim it's an arbitrary number that you made up, so why would google have an answer? Commented Dec 12, 2014 at 3:36

1 Answer 1


False acceptance results from the need to deliver a user friendly system and is tied to false rejection as well as the technical limitations of systems.

In short it is an accuracy issue.

A digital representation of a finger print (or any other biometric for that matter) is effectively a digital subset of the information about a finger print. The amount of information stored about each biometric record will be system dependent, but will be finite and therefore there is the potential for duplication. As with most digital representations some information will be lost and for an authnentication system there is a trade of in terms of storage space and processing speed which will limit the amount of information that can be used for the digital representation of each finger print.

As with any user system useability is key, with a biometric system users would soon stop using a systems if they had to submit their fingerprint repeatedly before it is accepted (false rejection), therefore the system has to have a tolerance threshold which allows for inaccuracies during the submission process by valid users i.e. a low false rejection rate, but decreasing false rejections will increase the possibility of a false acceptance because it is more likely that a different fingerprint will 'become' acceptable.

For your 1% FAR example, there is an expectation that out of 100 fingerprints it should be expected that 1 will result in a false match. Whether this means the person with the false match will always be able to create a false match is going to be dependent on how the system works (both hardware and software).


You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .