On cars, old, new and future; science & technology; vintage airplanes, computer flight simulation of them; Sherlockiana; our English language; travel; and other stuff
THE TECHNOLOGY of image recognition is as old as computers. To wit, consider the CAPTCHA test used to separate our recognition abilities from those of electronic machines.
CAPTCHA stands for “Completely Automated Public Turing Test to Tell Computers and Humans Apart.” Coined in 2000 by Luis von Ahn, Manuel Blum, Nicholas Hopper and John Langford at Carnegie Mellon University, CAPTCHA honors the memory of Alan Turing, the British logician and cryptologist who brought a philosophical view to the burgeoning era of automated computation. (See http://wp.me/p2ETap-a7.)
We find it straightforward, if not always trivial, to identify the CAPTCHA characters. Computers in their current state of development do not.
However, computer memories get larger and larger. And their image-recognition algorithms get better and better. The latest capabilities in facial recognition are an example of this.
The article “Never Forgetting a Face,” by Natasha Singer, in The New York Times, Sunday, May 18, 2014, got me thinking about character recognition in general, its good and its bad. For Singer’s analysis of the latest facial recognition, its societal pros and its cons, see http://goo.gl/DrwyQJ.
At their basis are physical attributes that add up to a recognizable face. Key measurements include the distance between the eyes, depth of eye sockets, width of nose, shape of cheekbones and length of jaw line.
The first facial recognition algorithms used two-dimensional imaging, with the subject looking directly at the camera, preferably in a controlled environment. (Think your DMV photo.) Even then, computer recognition has not been particularly successful.
Boston’s Logan Airport tried facial recognition with volunteers; it failed to find matches more than 60 percent of the time. See http://goo.gl/jmdy95 for background on facial recognition technology.
A major advancement is emerging with three-dimensional imaging that works with subjects in motion and at different view angles. Characteristics such as head size, contours and pose are combined with other basic biometrics. Also available are algorithms identifying skin textures, pore patterns and the like; in a sense, a skinprint.
To identify an individual, an image is compared with those in a data base. If the data base is a two-dimensional one, like the DMV’s, the three-dimensional image is first converted to the face-on view. In time and with increased computer power, it’s thought that collection of three-dimensional data could eliminate this requirement.
Like many aspects of technology, it’s fascinating—but also frightening from the point of view of personal privacy. A fingerprint, a DNA swab, even a computerized eye scan, requires an individual’s action, if not outright consent. Unlike these previous means of identification, advanced three-dimensional facial recognition is a passive one.
I’m entering a bank, waiting in line at airport check-in or walking through the mall, and my three-dimensional image might be captured for entry into an ID data base. In fact, it’s likely this has already occurred.
It is akin to—though rather more personal—than the photos of my home, its frontage and its backyard available from Google Maps.
All of these involve the concept of model release, an identifiable person granting permission to publish a photograph in one form or another. And then things get complicated indeed.
For instance, if it’s a news photo taken in a public place, no release is required. However, if the photo’s purpose is for trade or commercial use, then lack of a release could result in civil liability for whoever publishes the photo (note, not the photographer).
Do I own the image of my identity? At this point, it has been a question for the U.S. Court of Appeals for the Ninth Circuit, and possibly in time, the U.S. Supreme Court.
In the meantime, I’m going to keep a pleasant look on my face in public—or, for that matter, in private too. ds
© Dennis Simanaitis, SimanaitisSays.com, 2014