Comparison of Iris Recognition Algorithms Using Artificial Neural Network

Abstract

Iris recognition is regarded as the most reliable and accurate biometric identification system available. Iris recognition systems take an image of an individual's eye and the iris in the image is segmented and normalized for feature extraction process. The performance of iris recognition systems depend on segmentation and normalization, and segmentation succeeded depend on quality of image be captured, occurs error named segmentation error. In this research has reviewed iris recognition system and discussed two algorithms BP and BR to decrease error and determined what better than other about performance, error occur in non-iris region therefore, to process it can cut non-iris region from iris image during processing. Using artificial Neural network algorithms to process an iris image and using two algorithms and compare between its. Take a sample (captured images) , Design system model , segmented and extracted features of that images, using features as input to ANN technique, implemented some algorithms and compare between its using Matlab. The BPNN with different algorithms topology used in recognition because of its Recognition percentages for the irises tested images, but it needs more execution time for learning. the recognition rate of BPNN is more than the BR for the iris tested images, the size of the images are same in the both networks. find performance error in back propagation is bigger than in bayesian regularization. And time can be needed to designed model in bayesian regularization is more than in back propagation. in BP time to generate design is 1sec but in BR time equal 15sec, and error in BP 0.025872 but in BR equal 0.021009.