Biometric identification and facial recognition

CrunchGear reports on a new product called Lockface USB flash drive from Futen, a Japanese company. The flash drive uses facial recognition to identify its users. According to CrunchGear: “The first thing to do is to register a number of pictures of your face. After that, the Lockface verifies your face every time you need to access data on it (the verification process takes about a second). The USB drive doesn’t require extra software to be downloaded or installed. Alternatively, you can also use a password, completely ignoring the face recognition function of the device. It uses 256-bit AES to encrypt the data. Futen says the device has an error rate of about 2% (it verifies the “wrong” person in 1.91% of cases and won’t verify the right person in 1.98% of cases).”

Facial recognition software takes various features of a face, measures them and compares them to a list of parameters to form identification. Some of the parameters measured by facial recognition software include distance between the eyes, depth of the eye sockets, width of the nose, length of the jab line and shape of the cheekbones. Of course this works best when the facial recognition software has several different images of someone’s face from a variety of angles.

An internet search revealed accuracy rates of between 65% and 98% for facial recognition depending on the software, algorithm and process used. Sixty-five percent isn’t much to celebrate, but 98% is. The ability to control the environment in which the facial recognition software is used would certainly go along way in increasing accuracy. And that is exactly what a hospital is; a controlled environment.

I’ve always thought that facial recognition would be a nice feature to add to certain automated dispensing systems in the hospital; automated dispensing cabinets would be one technology that jumps to mind.