Abstract: The present paper presents an iris detection and localization method based on AdaBoost algorithm and neural networks with local interc onnection structures to overcome the limits in the current iris detection and localization algorithms. It has following features: according to the features of the iris image, we designed a set of local interconnection neural networks iris classifiers with different receptive fields and different complexity; we use d AdaBoost algorithm to integrate neural networks classifiers to generate a chief classifier with powerful iris detection ability; cascade connection structure was applied to increasing the detection speed. Experimental results show that this algorithm has a very high detection accuracy and speed. It can solve the iris detection and localization problems efficiently in the case of with large face region and those of cataract patients.

Key words: iris recognition, biometric recognition, iris detection, iris localization, neural network, AdaBoost algorithm