Iris Recognition Algorithm Research Of Images Acquired Under Natural Light

With the rapid development of digital and information-based society, intelligentidentification method based on biometrics are concerned by more and more people.Among the numerous biometric identification technology, iris recognition technology,because of its uniqueness, stability and collected non-invasive, has been widely usedin public security, information security, protection and other fields. But because of thelimitation of sampling device and pickup assembly, the iris recognition technologybased on near-infrared image already cannot satisfy security for identification in thefield of high reliability, efficiency and convenience. To make up for the shortage ofthe traditional iris recognition method in natural light environment, the irisrecognition method performing well in natural light is to research. Based on the studyof natural light of iris image recognition method for the target. And center for naturallight iris image acquisition, iris image preprocessing as well as the normalized irisimage feature extraction is conducted a series of problems such as discussed. Paper’smain work is as follows:1. Using the color correction algorithm for iris image under the natural lightcolor correction. Iris image under the natural light illumination, backgroundinfluence like image color deviation problem, and the problem of village inleads to deviation which occurs as the color feature extraction in this paper,the use of color correction algorithm can eliminate the effect good and toobtain more reliable color features.2. Iris localization method, which is different from traditional in the experimentfound that natural light within the outer edge of the iris image edge are muchmore clear, so this paper use the two-step fixing the iris outer edge, and indetermining the outside edge of the iris, using two step or activities withinthe circular template matching method can very good get locating on theedge of the results, this article is based on Hough transform in R channel foriris inner edge positioning and achieved good results.3. In feature extraction, considering the natural light to the color of the irisimage contains a large number of information, if convert it into grayscaleimage feature extraction can lead to color information of a large amount ofwaste, so, unlike traditional iris texture feature extraction, this paper proposesa color histogram feature extraction algorithm based on different color space,and UBIRIS.v2database and laboratory experiments on the acquisition of theiris database itself, and the commonly used iris recognition algorithm basedon Gabor filter comparison found that the proposed algorithm has betterperformance.