Research On Color Iris Localization Algorithm

With an emerging human biometrics for identification, color iris recognition has developed rapidly. Iris recognition technology has become one important branch in the cross-disciplinary of applied mathematics, image processing and pattern recognition. Comparing with gray-scale iris images, color iris images had acquisted more free and non-ideal, so obtained scientific and business widespread attention, and also had a better versatility and market prospect.Because research on color iris recognition is time limited, so algorithm is not maturity, there are lots of problems, it must discover new method to solve. In this paper, the main work is about color iris localization. It is one of the key problems in iris recognition that how to locate color iris’s position in an image.This paper presents a method of iris inner-outer boundary localization based on ellipse fitting. The main content in this paper is how to improve localization speed in the condition of ensuring the localization accuracy and stability.In this paper, the structure and working principle of iris recognition system are introduced firstly and current status of the gray-scale and color iris localization methods are elaborated. Furthermore, main problems about existing localization algorithms are analyzed and improved.Color iris localization algorithm was presented by analyzing the problem existed in the iris recognition system and the non-ideal color iris images’texture features. It consists of the following process, firstly, pupil’s location estimation; secondly, noise detection and process; thirdly, the iris outer boundary localization and upper-lower eyelid detection; finally, the iris inner boundary localization. Iris outer boundary localization uses a coarse-to-fine process, while in the outer localization the location of outer circle is estimated firstly, after which fine localization is performed. Localization is executed in the outspread image in polar coordinate, the method that circular boundary detection in the original image is transformed into detecting line boundary in the outspread image. As coarse result, the real boundary was detected based above detected boundary area. Then, ellipse fitting was adopted to improve the efficiency of the algorithm. Based on this condition, iris inner boundary was localized. Then the accuracy and robustness of iris localization are improved.The experiments on UBIRIS V2.0which contain lower quality color iris images such as defelcted, squinted and occluded show that the proposed algorithm is more accurate and stable, in ensuring the efficiency of the premise, which has been proved by experiments. Moreover, the correct rate of the iris localization is98.3%, it has a higher computational efficiency and better stability and versatility.