Research And Improvement Of Iris Recognition Technology Based On The Genetic Algorithm

Iris recognition technology has the advantage over other identity authentication technologies and is listed as one of the most safe and accurate identification technology because of the excellent biological properties of iris, which has a broad application prospect and important academic research value. However, because of the complexity of its application environment and the universality of its involved fields, its key technologies still need to be improved. This paper mainly focuses on a systematic analysis and research on the iris location, feature extraction, feature dimension reduction and other related technologies based on the properties of iris image and the common process of iris recognition system. The main work and contributions are provided as follows:Aiming at the flaws of the traditional Canny algorithm used in detecting the iris image edge information, a new improved Canny algorithm is proposed here after the study of the iris location segmentation model based on the combination of Canny operator and Hough transform. Firstly, Sober operator is adopted to calculate the gradient magnitude and direction of every single pixel point. Secondly, the process of non-maximum suppression according to the direction of the gradient is carried out in eight neighborhood. Finally, the method named Otsu is applied to set thresholds automatically, which makes the old Canny algorithm turn into an adaptive one. Combined the result with Hough transform to finish the localization of inner and outer edge of iris, which enhances the accuracy of positioning. At last, the iris region is asked to be normalized and enhanced.A feature vector which is produced by the 2D-Gabor filter has redundant information. In order to obtain a low dimension and efficient feature vector, the process of feature selection which is realized by an improved genetic algorithm is added here. Aiming at the flaws of the iris feature selection which is based on the genetic algorithm, an improved genetic algorithm combined with the advantages of particle swarm optimization algorithm is proposed here. Particle swarm optimization algorithm is integrated into the whole framework and adaptive genetic operators are designed. Finally, this paper takes a new Hamming distance as the principle to judge two iris images are belong to the same class or not. With the more efficient and lower dimension feature vector to represent iris information, the matching accuracy has been improved.The iris images tested in this paper is taken from the database named CASIA-V4-Thousand and CASIA-Iris-Lamp. The performance indexes named False Accept Rate?False Reject Rate?Correct Recognition Rate?Equal Error Rate and Receiver Operating Characteristic Curve are used to measure the performance of this improved iris system and the results verify the validity of the improved algorithms.