Liveness Detection

As any other authentication technique, fingerprint recognition is not totally spoof-proof. One of the potential threats for fingerprint-based systems is represented by the fake-finger attack. Recently, the feasibility of this kind of techniques has been highlighted by some researchers ([1], [2]): they showed how some commercial fingerprint devices can be spoofed by synthetic fingerprints developed with materials as gelatin, silicon and latex.

Finger-liveness and fake-finger detections are the most discussed disadvantages of fingerprint recognition with respect to other biometric modalities, since a secure and robust method to detect the liveness or truthfulness of a finger does not exist yet. This affects negatively the usage of fingerprint recognition as an authentication technique for very high secure applications.

In the Internet, it is very easy to find tools and materials to build a fake finger. The following picture shows some of the main steps to achieve this. You can find more information here.

As explained in the sensor section, many technologies to acquire fingerprints exist. Many fingerprint device vendors claim to have the final liveness detection approach, but after tests provided by third parties the methods always fail.

This video shows how it is possible to attack a fingerprint system.

(Well-known video of the Mythbusters was here)

Vulnerability of Touchless Fingerprinting

Touchless fingerprint technology is not free from vulnerability spoof. Indeed, the capture principle simplifies the method of attack. In the case of contact-based sensors, a fake fingerprint must replicate the 3D ridge-valley structure by the use of special materials (gelatin, silicon or latex). Instead, in the case of touchless devices, a 2D picture or a simple drawing of a fingerprint on a paper is sufficient to attach the system and grant an access.

A fake fingerprint can be easily obtained printing a ridge pattern on a matte colored paper. Then, the paper can be folded onto any finger and simply presented to the device. To make sure that the different material (paper) does not create wrong light reflection onto the camera to simulate more precisely the skin reflection properties, it is sufficient to use a paper with the same color of the illuminating light. This ensures to obtain a final image identical to the image obtained from a real finger.

This kind of tool was used to attach the device reported in [3]: a fingerprint image was printed on a green matte paper and then glued on to a finger. A commercial algorithm was used to match it against the image obtained from the real finger. The experiment was repeated for 20 fingerprints and the match was always positive.

To try to overcome this problem, new methods of protection must be investigated. In [4], the authors proposed a new method for liveness detection suitable for the touchless fingerprint technology. The method takes advantage of the sweating activity of the human body. Sweat drops come out periodically from the skin pores to maintain constant the body temperature. This activity is continuos and the frequency of the emission of the drops changes with the variation of the difference between the body and environmental temperatures.

Using high magnification lenses and special illumination techniques, it is possible to capture by the use of a remote camera the perspiration activity of the pores present on the friction ridges. The presence of this activity ensures the liveness of the finger and protects against a fake fingerprint attack as described above.

Experiments were conducted using a new generation high-definition camera (Sony HDR-H1C1E), optical-fiber illuminators and lenses with a magnification factor of 40x. The frame rate and the shutter speed of the camera were fixed to 30 frame/s and 1/250 ms, respectively. The acquired sweating-pores video sequences were then down-sampled to 1 frame/s to reduce computational load. Each frame was processed to extract the sweat pores using wavelets (top-hat wavelets) and traditional motion tracking techniques (optical-flow) were used to follow the presence/absence of sweat coming out from the pores present on the fingerprint friction ridges.

The proposed method is only suitable for touchless devices.

References

1.Matsumoto, T., Matsumoto, H., Yamada, K. and Hoshino, S. (2002) Impact of Artificial Gummy Fingers on Fingerprint Systems. Proceedings SPIE, Vol. 4677, pp. 275-289, San Jose, USA, Feb 2002.
2.Putte, T. and Keuning, J. (2000) Biometrical fingerprint recognition: Don't get your fingers burned. Proc. IFIP TC8/WG8.8, 4th Working Conf. Smart Card Research and Adv. App. pp. 289--303.
3.Diaz-Santana, E. and Parziale, G. (2006) Liveness Detection Method. Patent pending. EP06013258.6.
4.Parziale, G., Diaz-Santana, E. and Hauke, R. (2006) The Surround Imager: A Multi-camera Touchless Device to Acquire 3D Rolled-Equivalent Fingerprints, on Proc. of IAPR Int. Conf. on Biometrics, LNCS Vol. 3832, pp. 244-250, Hong Kong, 5-7 January 2006.