Fake Fingerprints - Liveness Detection Solutions

By Brian Rhodes, Published Oct 19, 2015, 12:00am EDT

One of the biggest concerns with fingerprint readers is how easy they can be fooled. While biometrics are typically more difficult to steal or fake, headlines still break news of fake fingers or stolen prints being used to fool sensors.

For this reason, many access control fingerprint readers include live finger or liveness detection that checks the finger being scanned is authentic.

In this note, we examine the four common methods (tissue reflection, heartbeat detection, dermal electric resistance, unnaturalness analysis), what HID, Morpho and Suprema use and why you need to beware of ambiguous claims.

Stealing ************ ** ***** **********

*** **** ***** ** the ******* ** **** while ************ *** ******, they *** ** ****** or **** ******* **********. Throughout *** *****, ******* methods ** ******** ******, by ***************** ******** ** ****** tips, ********** **** ***** *********, *****-**** ******* ** latent ****** [**** ** longer *********], ** **** using ******* ******* **** been ********.

***** *** ****** ** producing ******* ****'* ***** takes **** ****** **** stealing * ****, ***, or ***, *** **** is *** **** - unauthorized ****** **** **** access ** ********* ***** they ** *** ******.  

Four Common *******

***** *********** ****** ************* frequently *** ******** ********* methods, **** ** *** always ******* **** **** are ** *** **** work. ** *******, *** number *** **** ** methods * ************ ******* vary *** ********* **** into **** ********* **********:

  • ****** **********:*** **** ****** ****** (sometimes ****** ************* *******) typically **** ** ***** to ******* *** ********* contrast ** * ******'* skin. **** ****** ****** on *** **** **** normal, ****** **** ******** IR ***** ** * consistent *** **** ***** different ** **** ** covered ** ********* ********. Especially *** ******* ***** sensors, **** ***** ** done ** *** **** time ** *** *********** is '****', ** ***** is ** ***** ****** the ****. 
  • ********* *********: *** ** *** strongest ******* **** * high ******* ******** **** to ****** *** *********, rhythmic ******** ** *********** coursing **** *****. **** impulse *********** ** * beating *****, *** ******* it ******* *** ******** attempt ** *******.  ***** both ******* ** ******** and ******** ***** **** particular ****** *** ** the **** ********* ******* for ************* ** ***.
  • ****** ******** **********:*** ********** **** ** sensors, ******* ***** **** carries * ***** *** consistent ********** **********. ** a ****** ** ********* and *** ****** ** unable ** ******* ******* skin **********, ** ** invalidated. **** ** *** cost ********** ***** ******, this ******** ***** ** common, *** ** *** not ** ******** ** wet ** **** ************ that ****** *** ******* of **** *** ***** in ******.
  • ************* ********: **** ****** ***** ** the *******, ** ** relies ** ******** ****** alone ** ********* ************. This ****** ******** * print ******* ******* *************** of ***** ** ******* *******. ***** ** ****** checks **** *******, ****** or ***** *****, ***** print *****, ** ******** clarity ** *** *****, if *** ******* ** the **** ***** ******* a ******* '*********' *****, the ***** ** *********** as ****.

***** *** **** ****** of ******* ** *** market, * **** ******* Liveness ********* ******* $***, ***** * **** **** layers ******* ******* *** cost *** ****.  *** addition ** ******** ********* is **** *** ****** of ***** ***** **** drive ***** ******, ***** with ****** ****, *********** support, *** ************* *********** of *** ******.

Liveness ******* ****

** ****** *** **** of **** ** ******* prints, **** ********** *********** scanners *** ******* *** checks ** ******* **** are real.  ***** ** * list ** '******** *********' or '**** ****** *********' on ****** *********** ******* specsheets:

  • ********/*** ******: **** ********* ********* and ****** ********** ** validate ******* *** ****.
  • ******: ********* ** *** reader, ****** **** ****** reflection, ****** **********, ******* up **** ************* ********.
  • *******: **** * ****** of ************* ********** ** determine ** ****** *** faked ** *********** ** they *** ******.

** *******, ***** ********* implement **** **** * single ****** ** **** on *******- * *** point ** ******** *** wide ***** ** ********* fake/spoofed ***** ********.  

Beware ******** ******

*** *** '******** *********' methods *** ******* *********. For ******* *****'* ***** ** *** ****** *********** ****** (**** release) ** * ****** fingeprint *****. ***** ******* the ****** **** '******** detection ******** *** **********' (unnaturalness ********) ** ********** models. *******, ******* ***** ****** ****** effective *****, **** ***** ***** updates. ** *** **** of *****'* *********** ******, spoofed ** ***** ****** are ***** * **** with ******** ******* *****.

** *******, ********* ******* that *** ******** *** software **** *** ****** performing (**: *********, ****** Resistance, *** ****** **********). Take **** ** ***** methods ************* **** **** use, *** ** ********-**** (like ************* ********) ** unclear, ** ****.

Comments (9)

Thanks, Brian, for a really informative article.

It would be interesting to hear about hyperspectral's robustness to skin color variations. It seems reasonable that natural variation has been considered and accommodated, without substantially reducing ability to discriminate false presentations. In fact, it would be interesting (and probably more secure) if hyperspectral skin tones added another dimension to the fingerprint signature.

There must be a reasonably wide variation in acceptable conductivity to accommodate routinely varying conditions. Knowing that skin conductivity is relevant, how hard could it be to replicate a static resistance within a dummy finger, or even on a dummy image? Naively, considering that many laserjet inks are conductive, could you print a target resistance by varying print darkness? Who knows what is possible?

For heartbeat, one could print onto a flexible substrate (something like acetate), then glue it onto a bladder such as a balloon. While presenting that image to the reader, one could gently squeeze the bladder slightly more frequently than once per second.

If hyperspectral IR has a reasonable resolution, it seems as if it could be the most difficult to faithfully replicate, and that fingerprint augmentation with hyperspectral signature could even enhance security.

IMHO, one could obtain, at the expense of authorization delay, a higher level of confidence by using a challenge and response mechanism.

Something where the reader would display randomly changing "challenge", like a gesture, that would one mimic in response.

Of course, this is easier said than done, but could be useful since all static biometrics are, if someone is willing to spend enough $, vulnerable to the attack vector of 'fool the sensor'.

IMHO, it is too expensive (and usually affects accuracy) and can be solved by employing other methods. For instance, a couple of possible options:

  • enroll all fingers and request a different one each time or a combination
  • in the same price range it is more efficient (accurate, robust) to use palm vein or iris verification
  • the easiest (and probably the best) way is to use multi-factor authentication (PIN / Card / Fingerprint / Voice / Face)

[Request for Update (Suprema)]

Hi John / Brian,

Check this out: https://www.supremainc.com/en/AccessControl-TimeandAttendance/Biometric/BioStation-A2

Not all fingeprint devices from Suprema are using that advanced feature thought:

BioStation A2 is implementing advanced Live Finger Detection based on:

- Dynamic Pattern Analysis

- Unnaturalness Analysis

- Dual Source Light Imaging

BioStation A2 was tested and compared to nop notch fingerprint vendors as below:

#pushingthelimits

Best regards,

baud

Who would dig up grandma just to look at a bunch of military floppy discs?

Some people just need a hobby...

once again it seems like it comes down to adding layers to decrease risk. all about the how much the customer is willing to spend to protect their assets. 

I guess I should have read the next article before I asked the previous question.

Great article

Fantastic info.

Read this IPVM report for free.

This article is part of IPVM's 6,653 reports, 896 tests and is only available to members. To get a one-time preview of our work, enter your work email to access the full article.

Already a member? Login here | Join now
Loading Related Reports