Visible Wavelength Iris Segmentation: A Multi-Class Approach using Fully Convolutional Neuronal Networks

Iris segmentation under visible wavelengths (VWs) is a vital processing step for iris recognition systems operating ata-distance or in non-cooperative environments. In these scenarios the presence of various artefacts, e.g. occlusions or specular reflections, as well as out-of-focus blur represents a significant challenge. The vast majority of proposed iris segmentation algorithms under VW aim at discriminating the iris and noniris regions without taking into account the variability that is present in the non-iris region. In this paper, we introduce the idea of segmenting the iris region using a multi-class approach which differentiates additional classes, e.g. pupil or sclera, as opposed to commonly employed bi-class approaches (iris and non-iris). Experimental results conducted on two publicly available databases show that the use of the proposed multi-class approach improves the iris segmentation accuracy. Simultaneously, it also allows for the segmentation of different non-iris regions, e.g. glasses, which could be employed in further application scenarios.

I. Introduction

More than 20 years after the first iris recognition algorithm was developed by Daugman [1], iris biometrics remains an active and rapidly growing field of research. Among others, iris recognition based on visible wavelength (VW) images represents one major challenge. Independent tests [2], [3] have confirmed remarkable recognition accuracy for iris recognition for good quality near-infrared (NIR) images. If captured properly, VW iris images might exhibit a similar level of detail within the iris texture, while in general the amount of available information is expected to be less [4]. For instance, within VW iris images possible artefacts, such as specular reflections or shadows, are more pronounced, which generally leads to an increased intra-class variation. The segmentation of the iris involves a detection of inner and outer iris boundaries, a detection of eyelids, an exclusion of eyelashes as well as contact lense rings, and a scrubbing of specular reflections [5]. Accurate segmentation of the iris region in VW images represents one of the most critical tasks, since errors in the pre-processing stage significantly impact recognition accuracy [6].