The Research Of Preprocessing Localization Algorithm In Iris Recognition System

With the rapid development of Internet and IT technology, information security is becoming increasingly important, and people attach more and more importance to identification technology. In the rapid development of today’s society, the traditional identification technology has not fully meet the requirements of the people because of its inherent limitations. In such an environment, the biometric technology gradually attracts more attention. In recent years, biological recognition technology has become a hot topic in the field of authentication, with its advantages of stability, convenience and reliability, compared to traditional identification technology. As one kind of biometric technology, iris recognition is widely applied to security, mining, finance and other fields owning to its high degree of uniqueness, stability, security and anti-invasion. And Because of this, it has a broad development and application prospect.Iris recognition system consists mainly of iris image acquisition, preprocessing, feature extraction and matching identification in which iris image acquisition refers to the iris image acquisition by an effective image acquisition device. Iris image preprocessing usually refers to positioning iris image, detecting the eyelid and eyelashes, normalization operation and so on. Iris feature extraction means using certain methods to extract features in favor of matching identification from the image preprocessed. Matching identification uses some similarity measure to classify and recognize the features which have been obtained. Among them, the pre-processing is the key to iris recognition technology and pretreatment results will directly affect the accuracy of iris recognition. This paper mainly studies pre-processing iris localization algorithm.In the non-ideal iris image, iris image contains not only the iris region, but also eyelashes, eyelids, spot and other disturbances and these disturbances often affect the positioning of the iris and affect the accuracy of iris recognition system. Thus, good iris localization algorithm is necessary. Currently, iris localization algorithm which has high complexity and high demand for the image quality usually uses the entire iris image circle fitting, and the iris image orientation of non-ideal accuracy rate is not high. For non-ideal iris image, this paper proposes an iris location algorithm based on spot area and improved Hough transform. The algorithm mainly includes the following contents: (1) determining the reference point:using the feature of the pupil forming spot area because of reflecting to eliminate it with morphology operation. Then turning the iris image, before and after morphology operation, to binary image to locate the spot area and use the average spot area as a reference point(2) locating the iris inner boundary: using canny operator to identify the boundaries of the iris image, using effective reference point to determine the boundaries of the region fitting, the center, the parameter range of radius, and using improved Hough transform within these parameters to identify the iris inner boundary;(3) locating the iris outer boundary: changing the hysteresis threshold of canny operator to get outer boundary, then determining the effective region of the fitting outer boundary, the center and the parameter range of radius based on of the iris inner boundary. Similarly, using improved Hough transform within these parameters to identify the iris outer boundary.The iris localization algorithm proposed in this paper uses the iris image library of CASIAV3to carry out experiments. The experimental results show that the iris localization algorithm proposed in this paper can quickly and accurately locate the iris inner and outer boundaries.