Reliable Iris Detection by Boundary Search Using Haar-Like Features
Kazutaka Ohnuki, Tomohiko Ohtsuka, Hiroyuki Aoki · 2013
This paper proposes new reliable iris detection by boundary search based on Haar-like features, aiming to achieve high detection accuracy of iris region. The line search based on Haar-like features, which encode the existence of oriented contrasts in the iris image, is carried out in order to achieve stable detection of low contrast boundary between iris and sclera. After that, the restoration of iris outer boundary is carried out and it can achieve reliable detection even there are several obstacles, such as eyelids and eyelashes, because of the nature of Haar-like features. Experimental results obtained using the CASIA-IrisV3-Lamp show that the proposed approach can achieve the success detection ratio of 97.5%.