A novel phase-intensive local pattern for periocular recognition under visible spectrum;
Sambit Bakshi, P. K. Sa, B. Majhi; Department of CSE, NIT Rourkela
ris recognition with NIR images has been proven to be accurate and is already in
use in industry. However NIR image capture is costly and impossible in unconstrained
scenario. Iris recognition in visual spectrum, i.e. from images captured with regular
cameras, has been proven to yield very low accuracy. Hence this work
| attempts if recognition from visual spectrum eye image
can be achieved,if not by using iris pattern, but by using patterns in nearby eye regions (periocular).
However it is an independent challenge to answer the question, "How much region around eye is to be
considered for recognition?" We have considered 300% width of iris diameter as periocular template.
The article extracts a feature from this template based on its phase intensity. The proposed method
shows that extracting features and recognizing human is possible from periocular region even if the
image is captured with simple VS camera. More in Biocybernetics and Biomedical Engineering
(List of Research)