In December 2003 the BioSec project, co-funded by the VI Framework Programme of the European Commission (Biometrics & Security IST-2002-001766), started its two years life cycle with the purpose to consolidate and progress research on biometric technologies, and especially aliveness detection:
Within this project, I studied the possibility to detect fake fingers, using impedance with swipe-type fingerprint sensor such as the FingerChip. This is not so obvious because the finger is moving during acquisition, and so measurements must be very quick, and results may not be usable.
So we worked on different electrode shapes, associated with an impedancemeter able to acquire many measurements (magnitude + phase) per second at different frequencies.
The final demo was using the following packaging, with different electrodes to test:
We were using Windows 98 (to get rid off the driver problems to access the parallel port and the serial port, and remember, this was in 2004) and a specific "skin impedancemeter":
A specific program was able to acquire and display the collected data during a swipe, many (magnitude,phase) couples @ 10kHz, 100kHz & 500 kHz, as well as the acquired fingerprint:
The idea was to take advantage of the impedance response which is specific for the skin:
A score was computed to define if we were facing a real or a fake. Most fakes were easily rejected (majority of score at zero), but some of them -closer to some living material- were much more difficult to detect:
It may also happens that some real fingers were rejected from time to time, depending a lot on the chosen threshold (you may want to use 0.7 for high security, or 0.4 for a better user experience).
Here is an example of a real finger, accepted and recognized, and a fake, rejected but recognized:
This is an extremely low cost solution, as no additional sensor is required, just two electrodes! And able to work even with a moving finger. In 2008, the AT77C109 was integrating some fake finger detection means based on impedance.