First steps are promising, accuracy can be improved

Jun 11, 2015 13:44 GMT  ·  By

The National Institute of Standards and Technology (NIST) is working on finding the correct recipe for an algorithm that can distinguish automatically the bad guys from regular ones based on the body ink they sport.

The development is intended as an additional instrument for law enforcement, but an approved version could also be used to ascertain the identity of victims of natural disasters like earthquakes or tsunamis.

Examiner subjectivity is a problem

Among the challenges encountered by NIST with the existing tattoo recognition software was lack of consistency in human examiners labeling the same markings. The institute says that “the increasing variety of tattoo designs requires multiple keywords, and examiner subjectivity can lead to the same tattoo being labeled differently depending on the examiner.”

This issue determined NIST to challenge the industry and academia to start the development of an automated image-based tattoo matching technology, at the request of the FBI Biometric Center of Excellence (BCOE).

During a workshop (details and discussions here) on the matter, participants were asked to report their results on visually similar or related tattoos from different subjects, degradation of a tattoo on a subject over time, recognition of a mark based on a small part, similar or related tattoos starting from images (sketches, prints, digital or natural images), and detecting if an image includes a tattoo or not.

Better quality of collected images needed

The challenge was set up by NIST scientist Mei Ngan, who said that the areas that need more research are detection of visually similar tattoos on different people and recognizing a tattoo image from a sketch or sources other than a photo.

Collecting higher quality images is another thing that needs improved in order to increase recognition accuracy.

A total of six entities took part in the test: Compass Technical Consulting, the Fraunhofer Institute of Optronics, System Technologies and Image Exploitation, the French Alternative Energies and Atomic Energy Commission, MITRE, biometric technology firm MorphoTrak and Purdue University.