If the driver’s license or other personal data isn’t helping to catch a suspect, try their tattoo.
Anil Jain, a distinguished professor of computer science and engineering, and his research team have developed an automatic retrieval system to help law enforcement match tattoos to suspects.
The software system, called “Tattoo-ID,” stems from biometrics research, which involves working with people’s fingerprints and identification patterns, said Jain. The database goes a step further by linking tattoos, marks and scars to criminal records of all suspects.
Members of law enforcement will be able to provide a tattoo image inquiry, and the system will automatically retrieve the most similar matches from the database along with the criminal records that correspond with each suspect.
While Jain has been working with biometrics for the past 15 years, the specific type matching research related to tattoo identification began a year ago, he said. The team has been working for six or seven hours a week since then.
“Until recently, the search has been a pretty tedious process,” Jain said. “What we are doing is trying to find similarities between images rather than rely on all images from the database.”
The Michigan State Police has collaborated with Jain and his research team to provide access to their database of 100,000 digital images of tattoos already in existence, information that is usually only open to law enforcement officials, said Lt. T.J. Riegle of the Michigan State Police.
Jain’s system could prove to be very beneficial within the field, Riegle said.
“Once this database has been perfected, multiple departments will be interested,” he said. “Any types of marks that we can get to help us identify an individual is good in the long run for law enforcement.”
With Jain’s new system, an individual scrolling through the digital database of tattoos can further narrow down the list of suspects based on matching tattoos. While a tattoo is not as unique as a fingerprint, Jain said matching tattoos can help the search process for victims as well as suspects.
“Many times the police only find parts of the body, where fingerprints and face cannot be used,” Jain said. “You could then identify them based on their tattoo.”
The most challenging part of the research, Jain said, is determining how to define similarities between two images and “train” a database system to find similarities based on preference.
As well as identifying suspects and victims by tattoo, Jain and his team are working on matching scars and marks.
“The key advantage here is if you have a really large database, it’s often impossible to go through and find a match,” said Rong Jin, assistant professor of computer science and engineering and a member of Jain’s research team. “This system will facilitate the matching procedure and be able to identify a small subset.”
The system is currently a beta version, Jain said, but can be expected to be used in an operational setting in the next six months.
The goal is to make the system practical, Jin said.
“If we want to commercialize, they will make a substantial change for that,” he said.
Support student media!
Please consider donating to The State News and help fund the future of journalism.
Share and discuss “MSU team creates 'Tattoo-ID' system” on social media.