An Approach of Noisy Color Iris Segmentation Based on Hybrid Image Processing Techniques

摘要

Iris region extraction is the most controversial issue in the iris recognition system, since the poor results for this stage will spoil or break the effectiveness of the iris recognition system. In this paper a sophisticated system method to localize iris region is proposed by employing different image processing techniques: morphological operators (i.e., erosion and dilation), seed filling, rotation, 8 neighborhood operators. The proposed system is based on the formal attributes of an image of the iris, taking into consideration the noise area had been found in different parts of the eye image such as specular reflections, focus and small visible iris part. And the experiment was conducted using 1877 iris images of standard color data UBIRIS, and the result shows that the proposed method had a high rate of accuracy.