Detection of distorted 2-D iris data using multi-layered counter propagation neural network

Iris recognition is highly stable, accurate and secure verification technology for biometric authentication. The main focus of this paper, obtain a good identification percentage based on human iris. Further the image is successfully recognized when some pixels are missing or image is distorted. In the paper, counter propagation neural network is used for iris recognition. The CPN is a combination of unsupervised and supervised learning. The recognition rate of the proposed network is 99.6 %. The experiments are performed on eleven datasets with 1215 images it gives 99.6 % accuracy. The Performance of the Counter propagation neural network is compared with the Daughman´s method, Bole´s and Wildes traditional method. The network gives 5 and 6.84 percentage improvement over Daughman methods and 1.19 & 3.33 percentage improvement over Wildes methods´ for dataset A and dataset B respectively. The CPN method shows 4.22 & 6.89 percentage improvement over boles technique for two datasets respectively.