CSIR working on improving biometric ID techniques for minors
The Council for Scientific and Industrial Research (CSIR), an African research and development organization, is developing a biometric identification system that can obtain biometric data from children with greater accuracy.
In a new video, CSIR detailed the Department of Science and Technology-funded, two-year project in which researchers will be collecting ear, iris and fingerprint biometric samples.
Using this data, the organization aims to develop a system that can determine or verify the identities of minors throughout their childhood using reference biometric samples captured during their infancy.
CSIR researchers have identified the fingerprint, the iris, and the outer ear shape as suitable biometrics for the project.
The researchers will determine which of the three biometric modalities are best suited for the system.
Though the iris is the most unique and permanent biometric, there are a couple of issues to consider when acquiring iris scans from very young children, particularly infants. The greatest challenge is that infants are often sleeping and therefore have their eyes closed.
“The ear is a new avenue that the CSIR is looking into. The pattern or external shape of the ear is fairly consistent as the child grows,” said Kribashnee Dorasamy, researcher at CSIR. “We will be looking into 2D to 3D matching. The 3D captioning allows us to extract more information about the ear itself.
“For the capturing of the ear itself, we use a 3D scanner and the scanner basically projects patterns on the ear and this captures multiple scans. These scans are stitched together to create a full 3D representation of the ear. To give us more flexibility around the head, we created a mock motorized mount that will move the 3D scanner and capture multiple scans automatically where we can stitch the image to create the full 3D representation of the ear.”
Fingerprints also have their share of challenges when it comes to identifying children. Although fingerprint patterns remain constant over time, the relative pattern itself changes in scale as the child grows.
“Existing technology did not take that into account when it was being designed because it was designed particularly for adults,” said Tshegofatso Thejane, a researcher at CSIR. “So we as the CSIR are looking to develop technologies and algorithms that can at first, find a better way of acquiring fingerprints from children, and also, model the growth of a child such that the scaling issue does not affect the accuracy that we get.”
Last month at ID4Africa, Anil Jain, professor of computer science at Michigan State University, revealed study results that show fingerprints can be used to effectively verify the identity of children enrolled as young as six months of age.