4.6 False identification
If you think back to the Heathrow experimental system and the United Arab Emirates system described earlier, you can see that the false matches and false non-matches open up possibilities for these systems to malfunction. In the Heathrow scheme, a false match could mean that a person who was not enrolled might be allowed through. In the United Arab Emirates scheme, a false non-match might mean that a person who should be stopped is allowed through. These are examples of identification error.
Work through the animation below, then move on to the next Activity. (Click on the link below to view in a separate window.)
Work your way through the animation above called Identification and verification. It will take around 15 minutes. The animation reviews the material you have already studied on identifiation, and introduces the concept of verification which you will study shortly.
The animation shows that the rate of false positive identification depends on two things: the false match rate and the size of the database. For example, if a system has a false match rate of 1 in 1000, when used on a database of 1000 (or bigger) it is virtually certain that there will be a false positive identification with someone in the database. Thus, the false positive identification rate in this case approaches 100 per cent (1 in 1), because of the large size of the database. A false positive identification rate of 100 per cent means that every time a sample is checked against an entire database, there is at least one false match somewhere in the comparisons.
Besides reviewing the idea of identification, this has introduced the concept of verification, which we shall come to shortly.
To summarise, false positive identification arises from a false match with a template belonging to someone enrolled in a database, as shown in Figure 9.
The rate at which this happens depends on both the false match rate and the size of the database.
Note that it is possible for a piece of biometric data to falsely match more than one template in the database. It is also possible for a piece of biometric data to correctly match one template, and to falsely match one or more other templates.
We have probably all experienced false positive identification error when we see someone we think we know, say 'Hello' to them, and then realise it is not the person we thought it was. Something about them looked familiar, but the similarity was deceptive.
False negative identification error arises from failure to match the person with their own template in a database (Figure 10).
When there is false negative identification, it is conceivable that the person's data could falsely match someone else's template. However, this is irrelevant to the meaning of false negative identification, which is strictly about the failure of a person's data to match their own template.
We have probably all experienced false negative identification when we fail to recognise someone we already know, possibly because they have changed their appearance or because we meet them in an unfamiliar context.