Improved Methods Match Suspect Sketches to Mugshots

(Michigan State Univ.)

A computer science/engineering team at Michigan State University in East Lansing has developed a set of algorithms and created software that automatically match hand-drawn facial sketches to mug shots stored in law enforcement databases. Their findings appear in the March 2011 issue of the journal IEEE Transactions on Pattern Analysis and Machine Intelligence (paid subscription required).

The long-time practice of using police facial sketches to catch criminals has been, at best, an inexact art. Typically, sketches are drawn by artists from information obtained from a witness, which are not always accurate depictions of the suspects. And the few software programs available that produce sketches based on a witnesses descriptions tend to be less accurate than sketches drawn by a trained forensic artist.

The team led by MSU professor of computer science and engineering Anil Jain and doctoral student Brendan Klare undertook what the university calls the first large-scale experiment matching operational forensic sketches with photographs. The researchers matched up high-level features from both sketches and photos — features such as the structural distribution and the shape of the eyes, nose and chin. All of the sketches used were from real crimes where the criminal was later identified.

So far, say the researchers, the results have been promising. “We improved significantly on one of the top commercial face-recognition systems,” Klare notes. “Using a database of more than 10,000 mug shot photos, 45 percent of the time we had the correct person.” The researchers plan to field test the system in about a year.

*     *     *