Research On Iris Localization Algorithms

With the social development, we are becoming increasingly more dependent on the identification and authentication means. The traditional authentication technologies like password mode cannot meet all the requirements for being often restricted to use in certain scenarios. Thus biometric identification is proposed, which uses human biological feature as main identification media. The paper is rightly to focus on iris recognition research, improve iris localization methods and propose a framework for iris recognition system.Iris localization, one step of the iris localization processes, is to localize and extract the iris region from eye image. Now, iris localization is one of the most important branches researched in iris recognition.Further development and applications of iris localization algorithms is currently restricted by some shortcomings such as high computational complexity, long time cost, and poor robustness, etc. Based on current two-phase iris localization algorithms, an improved three-phase iris localization method is proposed. We abandon some images by carrying out the quality of the iris image, so as to get an increasing accuracy and anti-noise ability. Firstly, the center and radius value of the inner boundary is estimated. Then iris position is calculated for the second time and finally we get the exact pupil position. Finally, the outer boundary is found using results of calculation. The comparison experiment shows the improved method can reduce the time cost and increase the accuracy, especially for the poor-quality images with eyebrow, eyelash or eyelid overlapped. Since the inner and outer boundaries are carried out, we detect the regions of pupil, sclera, eyelids, eyebrows and any other noise pixels, remove them and normalize the pure iris image for extracting and matching the iris feature.The paper also designs a framework of iris recognition system using the iris recognition algorithms, which consists of three parts including core algorithm module, embedded devices and an iris information management system. The framework supports Linux and ARM too. Among the framework, the management system deals with personnel and iris information, real-time and off-time iris feature capture, also can run duplex communication with mobile devices to transfer feature codes.