Research on iris pattern matching method based zero-crossing wavelet transform

Sharp variation points are among the most meaningful features for characterizing transient signals. As a result, The Sharp variation points can be detected by zero-crossing, and they have been selected as the meaningful features for iris recognition. It is less sensitive to the influence of noise and illumination. But the numbers of zero-crossing points is not fixed, which is unlike the same length of iris feature codes as Daugman proposed, so it will be difficult to match. Boles used the dissimilarity function to solve this question, but it need long time to compute the dissimilarity values. For Huasdorff distance reflects the most mismatch extent of two sets and it will not require the two sets have same length elements, we used the modified partial hausdorff distance to match the feature points. In the experiment, the iris images were corrupted with varying degrees of white Gaussian noise to find the effect about noise. The result showed the method was less sensitive to the change of noise. In additional, we compared the modified partial hausdorff distance matching method with the dissimilarity function method and found that the modified partial hausdorff distance matching method's computing time is much less than that of the dissimilarity function method. Under experiment circumstance, the matching method performed effectively and had a better robust.