The Research On Low-quality Iris Image Registration

As soon as the technology of image registration has more matured,the technology of image registration has been applied to the field of automatic target recognition,medical image processing Intelligent robot,intelligent manufacturing,military applications and so on,and gained a great deal of research results.The iris,which as a very important biological feature, has more valuable applications in the field of Biometrics.The identification of the iris is accurate,stable,and difficult to be counterfeited or stolen,it is much better than other biological features.The experiments show that the high recognition rate of iris is based on high-quality iris images,and low-quality iris(for example,is covered by the eyelashes and eyelids,contains noise or low-resolution) will affect the accuracy of iris recognition seriously. In these circumstances,the acquisition of the iris image is covered by eyelashes and eyelid is a more common phenomenon.The traditional processing of the iris image is only choose the 60 degrees fan-shaped region from the iris as the research object,it is obviously that the traditional approach discards a great deal of useful iris information.In this paper,we have proposed the idea that applying the technology of image registration in the field of low-quality iris images.The goal is to increase the rate of iris recognition in the condition of not discard the useful iris information as far as possible.The traditional method of image registration preprocess focused on the rigid object.The Image registration has two major methods:the first is based on the bright,and the other is based on the feature.The second method included three ideas,based on the feature points,the feature edges and the feature moments.In this paper we apply the Harris described one of the based on the feature points.The difference between the iris image registration and the natural image registration is that the iris image is non-rigid,and the deformation of iris is very irregular.Thus the traditional method of image registration to find a transform matrix does not very effective here.In this paper we have proposed that apply curve fitting technology in iris registration.