At the core of a biometric system is the one-to-many identification process, where the input sample is compared with the stored templates in the biometric database. It is considered a “positive match” when the comparison of the input sample and a stored template is above a pre-defined threshold. When the system mistakenly matches an input sample with a stored template of someone else, it is a “false match.” So, when a system is said to have one in a million chance of a “false match,” how good is the system really?
Let’s say you have 10,000 people enrolled in a biometric system. Each time, every person’s sample is compared against the 10,000 templates stored in the database. In theory, there are 10,000 X 10,000 = 100,000,000 comparisons to check all 10,000 people. After removing duplicates and the comparison against one’s own template, there are still (10,000 X 9,999) / 2 = 49,995,000 “unique” comparisons. This means that even if the odds of “false match” are one in a million, there are about 50 chances that the system would mistakenly match a person as someone else with a population of only 10,000 people.
So when you have a biometric system with a one in a million chance of a “false match,” think again about how good it really is within your environment.