Research Of Iris Recognition Based On Mathematical Morphology And Rough Sets

Personal identification has played an increasing significant role in today's society of juridification and informationization, but traditional identification methods constantly give the oppertunity for criminals to violate the law. So, the personal identification technology based on the biometric character technology, such as fingerprint or palm reader, handwriting identification, gait identification and DNA test, has been paid attention by more and more researchers. In the current methods of biometric identification, the recognition based on iris has the best stability, accuracy and safety through the comprehensive evaluation of various performance indicators. In consequence, the research and development of iris recognition system has become a hot topic at the present stage.Although there have been some products about iris recognition in the market, the technology of image acquisition and algorithm research has the potential to further tap. So the researchers in various countries try to use different methods to improve recognition accuracy and efficiency. Accordingly, the current products of iris recognition systems are constantly upgrading for its higher using value and more convient application, which is the meaning of iris recognition research.The main work of this paper is a study of iris recognition algorithms, which can use new theory and propose new point of views based on the traditional algorithms. The characteristic used by iris identification is the internal texture, and the key is the speckles of the texture, which are considered as characteristics regions in this paper. Then encode and match the iris images though extracting characteristic points of the regions.In the stage of iris image preprocessing, the positioning of the iris is an important step. The method proposed in this paper is combined of rough positioning and pointwise detection in a limited area can correctly find out the inner and the outer boundary of the iris, simultaneously calculating the parameters of the boundary circles. This method is a creation in this paper for its simpty and efficiency. The eyelids, eyelashes, and lightspot produced by shooting are the image noises when iris texture is extracted, and various ways of mathematical morphology methods in image processing is used to deal with these noises, which have achieved satisfied treatment effect. For gaining the speckles of iris, the images should be segmented by watershed algorithms because for its well-behaved in grayscale image segmention. We can still effectively acquire characteristic values of the speckles, and encode through the location of the center of them. In this paper, the number, size and center of the iris' speckle regions are considered as the properties of iris texture. Rough set theory is introduced in the progeess of image matching. The matching efficiency will be improved greatly if those pictures which are not bound to be registered can be shut up. So the calculated quantities can be reduced before matching the codes one-by-one bit. The use of this approach is another innovation point of this paper.After simulation tests, it is showed that the proposed method is convenient and feasible. And the method has already met the basic requirements and can complete the task about identification although it is not perfect.