A face-recognition system that can see who you are even if you're wearing a face mask covering nose and mouth is discussed in the International Journal of Computational Vision and Robotics. The new system uses biometrics from around the eyes and forehead.
The research addresses a weakness in face recognition as a technology that was exposed during the pandemic. During that period when masks were often compulsory, their use caused conventional face-recognition systems to reject some authorised users. Unlike iris recognition, which can require specialised near-infrared imaging and controlled conditions, periocular recognition can operate with ordinary visible-light cameras.
The team has combined several types of visual information. First, a deep-learning model extracts features from around the eyes. Secondly, two image-processing techniques, local binary patterns and histograms of orientated gradients, analyse the person's forehead. The approach then combines the data to identify or verify an individual with an accuracy up to about 96 per cent in tests on one database but rather less on other testbeds.
Given that, the approach might be used as an initial screening step rather than as definitive proof of identity. A facial match could prompt further authentication with a second factor such as a password, PIN or security token. However, the team explains that they should be able to add other feature recognition into the same system, ears, face shape or profile, and those should improve performance significantly.
Agarwal, D. and Bansal, A. (2026) 'Fusion of periocular and forehead features for masked face recognition', Int. J. Computational Vision and Robotics, Vol. 17, No. 2, pp.133-158.
News media may use this press release as source material, in whole or in part, provided the content is not materially misrepresented. A link back to the original article is appreciated.
No comments:
Post a Comment