A research team from the University of Twente’s Services, Cybersecurity and Safety department has developed a new system for forensic facial recognition. The Dutch researchers outlined their method in the IET Biometrics academic journal.
Basically, the system enables the construction of a virtual, 3D face based on multiple still images of a face from various angles. As the researchers outline, it’s based on the Lambert reflection model, and doesn’t require a base of facial feature data for its digital reconstruction. This means that the modelling system helps to remove bias, using only actual image data to construct the face. As such, it could be extremely useful as a forensic method in the courtroom.
It could also be useful to a number of police and security agencies for their own investigations. The use of facial recognition with CCTV surveillance footage has been increasingly popular in the security sector, and various related government agencies are always looking for better means of biometric identification with respect to still images and other archival data. Along with other prominent academies exploring systems that can predict wrinkles on the face or match real faces to police sketches, the Twente researchers’ contribution could lead to a vastly improved identification capability for a range of interests.
Source: Phys.org
—
January 21, 2016 – by Alex Perala
Follow Us