Schneier on Security
A blog covering security and security technology.
« Giveaway: Liars and Outliers Galleys |
| Me on Airport Security »
December 23, 2011
Human Ear Biometric
I have no idea how good this biometric actually is.
Posted on December 23, 2011 at 7:51 AM
• 17 Comments
To receive these entries once a month by e-mail, sign up for the Crypto-Gram Newsletter.
If you do any sort of wrestling sport, this biometric is utterly useless. Ears tend to suffer in this sort of sport.
In general, this should be good. However, it is not difficult to totally change the ear's configuration with some simple plastic surgery, or even a prosthetic.
Jiu-Jitsu fighters would have problems with that biometric.
There has actually been a Dutch forensics expert working on ear biometrics for years. Also assisting in FBI cases. I don't think it's convenient for passports etc, but in burglary cases it has been very useful. Burglar tend to put there ears on windows and other surfaces to listen for the presence of people.
So yes, having a wrestler rip your ears off would make it difficult to be identified, but burglars don't tend to do that..
The two main issues of a Bio-metric are,
1, How stable it is.
2, How easy/reliable it is to measure.
As noted above ears are not very stable in everyday life, they get damaged very easily and changing some of their features can be done at home with out much pain or difficulty. Also as men get older their ears tend to change quite significantly from their 40's onwards. So in effect ears are not as stable as many other biometrics.
Then how do you measure them... they are in effect soft tisue with a cartilage inner, as we know both are sensitive to disease and certain chemicals, poor diet etc, so you would have to find "fixed points" that can be used as refrence points to make the other measurments from. Unlike cheekbones eye sockets and other bone refrence points nearly all the ear features can change or move simply by wearing a hat or scarf or be covered up by them and hair etc. Thus the ears are not often going to be visable to measure reliably without "their owner" being aware of it (for instance my ears are nearly always covered up by my hair, or a hat etc) by being "commanded" to make them available.
But yes they can be used but not as reliably or as well as other bio-metrics, so the question then becomes can they be used more quickly and easily than the more reliable bio-metrics to act as some kind of "pre-filter" etc.
@ Spaceman Spiff
or even a boxing match with Mike Tyson.
> "But yes they can be used but not as reliably or as well as other bio-metrics, so the question then becomes can they be used more quickly and easily than the more reliable bio-metrics to act as some kind of "pre-filter" etc."
Everything you pointed out points type 2 errors, and many of them are not day to day issues - the ears do not grow at a rate that will be a problem on a daily basis. So long as the system can slowly reflect those changes over time, and the user doesn't authenticate once a year, that should be no issue.
And yes, the user is going to be aware of the biometrics - this seems something that would be overt, but it is no different than eyes, which glasses (much more common than scarfs) can mess up (and as my daughter seems to demonstrate, hair can be an issue as well).
What this sounds like is a reliable enough overt authentication that is a bit less invasive than the various forms optical scanning.
I think it's saying the best they can manage at the moment is a bit under 90% accuracy with the best conditions (undamaged ear, good reference picture, and so forth).
Considering how many popular forensic techniques have been blown apart by being subjected to scientific analysis in recent years, I'd want to see a rigorous examination of "ear prints" before I believe that's actually helping...
It seems to me that the Brit's discovered a Hitler stand-in because his ear was different from Hitler's
Have you all actually read the conclusion: "In this paper, we attempted to analyze the symmetry of the two ears".
Resulting in "...indicate that left and right ears are symmetric only to some extent..."
So their research led to the answer "No". Yes?
But in true academic style: "Future plans include designing parametric models for the ear edges and using them to quantify the symmetry of
individual ear parts."
So if you give them more money they will come up with more unuseful results...:-)
This may be the giveaway: "The data collection process was funded by the FBI
Academy." Which make it all OK I guess...
Bumometrics, biometrics, earmetrics, isn't it all the same as sending a digital picture of your ass? Why not have seats read your bumometric thermal map? Isn't it just sending a digital representation of a pic? I fail to see how any of it is even remotely secure.
The idea of theater seats being able to recognize which ass should be in them is vaguely amusing, especially if the wrong one gets a mild shock. Butt really...
I'm not any kind of sec guru or anything, but the little bit of biometrics I'd seen a while back didn't seem all that great then either. But it struck me that a lot of people vending security tech are doing so with the understanding that their thingamajig could be just one component in the general idea of 'layering' filters of various sorts that protect 'X' stuff.
So the problem with the glut of biometrics seems to me that in practical application, that industry hopes to sell things that require individuals to get just about every body part that can be scanned, to be scanned. Wether that's before or after various other authentication methods is up to the people who think 'X' should be protected.
I think menacing looking people pointing guns at a bad guy trying to get to 'X' would still be a lot more intimidating than a little machine you stuff your finger, ear, eye, etc in, don't you? If you were the tech savvy bad guys, which would you be more worried about: bullets or biometrics?
In the UK I believe ear prints were found to be unreliable forensic evidence and new regulations came about because of this.
Addendum to my above comment: The UK Law Commission determined that earprints were unsuitable to be used, and this prompted more rigorous scrutiny of accepted forensic practices.
See Wilson's journal article for more info.
So, what do people think about this Ed Chiarini character at Wellaware1.com who has all this dodgy evidence of actors playing people in politics? He uses Ear Biometrics has his only proof and the photos aren't always that great, and any differences in ear shapes he puts down to cosmetics.. Is this method really that reliable?
Schneier.com is a personal website. Opinions expressed are not necessarily those of BT.