The Research Of Iris Recognition Preprocessing Algorithm

Iris identification has become applied mathematics and pattern recognition interdisciplinary hot research topic. Iris recognition preprocessing is one of the most important content which influences the correctness of iris outer boundary location, inner boundary localization, feature extraction and matching directly in iris recogn-ition.This article discusses the algorithm of iris recognition preprocessing. It includes iris image quality evaluation, locating iris area, iris noise detecting and prepares for feature extraction, matching. The main tasks of this thesis are as follows:(1)Iris image quality evaluationThe aim of iris image quality evaluation is that remove unqualified iris images, improve the accuracy of iris recognition. Unqualified iris images include rocing, the area of pupil is so little, iris outer boundary is not complete. In this article, the algorit-hmm includes threshold, morphology, and ellipse fitting.(2) Iris region detectionThe aim of iris effective area detection is to improve the efficiency of iris recognition algorithm. This section adopts algorithm of fuzzy f-means flustering algorithm(fcm) and Hough transform combining with each other. This algorithm increases the step of searching circle center, reduce the whole algorithm time.(3) Iris noise detectingThrough research of iris noise, this article proposed noise detection as follows: Gradient combining with gray were used to detect the spot then fills it with linear interpolation, Gabor filter were used to detect eyelash, and RANSAC were used to choose boundary points to fit eyelid.The algorithms put forward in this thesis have been validated by a plenty of simulating tests and experiments on database CASIA-IrisV3-Lamp and UBIRIS.v2, the results of NICE.II show that these algorithms are feasible to make precise preprocessing iris images.