How well do the various biometric authentication methods currently perform in regards to precision and recall? Can anyone cite a relevant research article?
Though this is not measured science with hard figures, it is known that the "security performance" of the iPhone fingerprint reader sucks.
Now the question might be the wrong one. A scanner is a scanner; it takes a "picture" of the finger (the iPhone scanner is supposed to use some longer-wave emission which very slightly penetrate into the skin, but that's still a kind of picture). What makes a "good" fingerprint scanner is its resolution, and also whether it will also measure temperature, blood pressure and other such data elements which can be strong indicators of an abnormal situation (fake finger; real finger no longer attached to a living body; finger owner is under heavy stress;...). As a rule, phone readers will offer decent resolution but nothing more.
Yet the most visible "performance" indicators of a fingerprint-reader system are the false negative/positive ratios:
- How many good fingers are unduly rejected
- How many bad fingers are incorrectly accepted
Beyond the scanner resolution, nothing in the scanner impacts these values. This is all a question of processing of the picture, and that's not done in the reader itself.