An Accurate and Robust Method Toward Iris Segmentation

The iris segmentation defines the effective image region used for subsequent processing, such as feature extraction and matching of the iris. This chapter proposes a novel method in iris segmentation. It proposes an effective method to get adaptive threshold to obtain the corresponding binary image. To remove the negative effectively from noise, such as hairs, it takes the advantage of the pots that are reflected in the pupil when people stare at the camera. The binary image is divided into several blocks and the number of black pixels in each block is calculated. The block containing pots and having the highest number of black pixels is the candidate region. The center of the darkest block is the key point as one point in the pupil. The pots are filled with neighboring pixels and grads play a key role in detecting the boundary points. To classify the possible boundary points, the neighbor function criterion algorithm originated from the pattern recognition is adopted and least square is used for curve fitting. The results on the challenging iris databases, such as CASIA-IrisV3-Twins, demonstrate that the algorithm is excellent in both accuracy and speed.