Research On Iris Recognition Technology For Personal Identification

Biometric identification technology has merits of no memory, nicely security performance and available anytime anywhere. As the novel biometrics, iris recognition technology has advantages of unique, stable, non-invasive and anti-counterfeit. Compared with other biometric technologies, the iris recognition technology has better correct recognition rate,so it attracts extensive attention,has wide prospect in market and scientific research value.This paper introduces the current research status of iris identification technology and the component of iris recognition system. Based on the existing iris identification technology, some of the key issues related to iris recognition system are investigated.In preprocessing of iris image, an iris circle fitting location method based on spring model is used, which can achieve iris location quickly and accurately. Fitting iris boundaries with circle has deviation in the accuracy of location, so a method based on varitional level set is proposed to achieve iris boundaries location accurately. In addition, according to the characteristics of the eyelids, an improved Canny iris algorithm is proposed to detect eyelids, which can reduce distraction of non-eyelid edge, extract eyelids edge selectively and achieve eyelid detection quickly and effectively.In iris recognition, two recognition methods are used. One iris recognition method based on feature extraction of two dimension symmetrical Gabor. In matching stage, hamming distances matching method and standard deviation of offset hamming distances are used, experimental results show that the standard deviation of offset hamming distances achieves better recognition effect. In addition, in order to decreases the complexity of iris recognition systems, this paper presents a method of iris image recognition based on regions of selected and global features that extracted without normalized the preprocessed iris. The method uses a bank of non-tensor product wavelet filters to obtain iris global features and scale invariant feature transform (SIFT) method to capture local features of selected regions. Then the similarity distances of local and global features are measured with different weights. This method can decrease the complexity of iris recognition systems. experimental results show that the method obtains good recognition performance.