It is amazing how quickly you get used to a car that starts as long as you have the key somewhere on your person. When you switch vehicles, it becomes a nuisance to fish the key out and insert it into the ignition. Biometrics aims to make it even easier. Why carry around a key (or an access card), if a computer can uniquely identify you?
[Alexis Ospitia] wanted to experiment with vein matching biometrics and had good results with a Raspberry Pi, a web cam, and a custom IR illumination system. Apparently, hemoglobin is a good IR reflector and the pattern of veins in your hand is as unique as other biometrics (like fingerprints, ear prints, and retina vein patterns). [Alexis’] post is in Spanish, but Google Translate does a fine job as soon as you realize that it thinks “fingerprint” is “footprint.” The software uses OpenCV, but we’ve seen the same thing done in MATLAB (see the video below).
Fingerprint scanners looked promising, but there are concerns about them. They are easy to spoof and if they aren’t sophisticated enough, they are subject to a literal hack (we don’t want someone lurking around the ATM machine with a pair of bolt cutters). We figure a severed hand won’t have enough blood in it to fool this system, but we aren’t willing to test that theory.
There are commercial systems that use similar technology from Fijitsu, Toshiba, and others. We have to wonder if there are other places you might have unique vein patterns that would be useful to scan. We’ve talked about fingerprint scanners before, and while we’ve covered someone spoofing a hand scanner, we don’t think that method would work with this particular setup.
Awesome project! I’ve had a NFC chip implanted in my hand about 1.5 months ago, its very basic, but allows me to store some information in my hand, and I made it so that I can unlock my PC by scanning my hand with a basic NFC reader.
Using this technique is not new and in fact I was using this technology in 1997 and earlier for a Client that was doing some specialty work for the US and CAN governments. In our case we used a type of “fingerprint” scanner that actually mapped out the blood vessel patterns beneath the surface of the skin. The technology worked extremely well and in fact there was no known false acceptance ever recorded. The only issues were that the user had to have adequate blood circulation to their fingertips (i.e. people with Raynaud’s syndrome had issues) and suitable mounting space (the reader was a bit big because of the optics within). Since the system was mapping active blood vessel patterns, the user had to be alive and the finger still attached. Since we could easily distinguish between fingers, we added a “duress” fingerprint. This was done so that if a user was under gunpoint, etc they would simply use their “duress” finger and the system would then take appropriate action (i.e. dispatch security personnel, create a false access portal, etc).
Very cool stuff and interestingly this was done almost 20 years ago!
Combine that with a latch and pain box to catch unauthorized users and you’ll have a better system. :)
Not already confirmed, are you sure Sure ?
The reason why google thinks ‘Footprints” is because in Spanish you can just say “prints” instead of the longer “fingerprints” or “footprints” so the google translator seeing just “prints” goes with “footprints” as default and to be fair without context a person would do the same. The difference is a person takes context into account.
“We have to wonder if there are other places you might have unique vein patterns that would be useful to scan.” Wiener, but I am not using THAT scanner in public.
Is the source code available anywhere? The instructable only contains a description of the hardware setup and a block diagram of the OpenCV image processing. However, I did not find any source code for the image analysis. Implementing that with both a low false positive and a low false negative rate is probably the biggest challenge in creating a practically usable biometric system.