Intelligent new software matches police sketches to mug shots

first_imgMatching police sketches to mug shots has been a tactic used to catch criminals for decades. Still, many criminals remain on the loose because the victim or witness’ description given to the sketch artist just doesn’t produce the same look as in the mug shot. We live in such a technologically-rich time that you’d think there would be a better, more accurate way to match these two pictures together to produce a successful result.A team led by Michigan State University Distinguished Professor of Computer Science and Engineering, Anil Jain and doctoral student Brendan Klare, have been working on just that. They’ve created software that would cross-match police sketches against real mug shots that are already in the law enforcement databases. Jain said they don’t match the photos pixel by pixel. The algorithms find high-level features from both the sketch and the photo, such as the structural distribution and the shape of the eyes, nose, and chin.According to a press release, their project is the first large-scale experiment to link sketches and mug shots, and the results have been promising. Klare said:We improved significantly on one of the top commercial face-recognition systems. Using a database of more than 10,000 mug shot photos, 45 percent of the time we had the correct person. True, 45 percent is an “F-minus” if we were to give this MSU team a grade, but it’s still early in the game, and the team plans to do field testing in about a year.As mentioned above, you think there’d be some sort of program or software to make this whole process easier. According to the release, there are a few commercial software programs available that produce sketches based on descriptions, but these programs are often even less accurate than those drawn by a trained forensic artist.Jain and Klare’s research is being published in this month’s addition of the journal IEEE Transactions on Pattern Analysis and Machine Intelligence.Read more at Michigan State Universitylast_img

Leave a Reply

Your email address will not be published. Required fields are marked *