W.: Boosting ordinal features for accurate and fast iris recognition

In this paper, we present a novel iris recognition method based on learned ordinal features.Firstly, taking full advan-tages of the properties of iris textures, a new iris representa-tion method based on regional ordinal measure encoding is presented, which provides an over-complete iris feature set for learning. Secondly, a novel Similarity Oriented Boost-ing (SOBoost) algorithm is proposed to train an efficient and stable classifier with a small set of features. Com-pared with Adaboost, SOBoost is advantageous in that it operates on similarity oriented training samples, and there-fore provides a better way for boosting strong classifiers. Finally, the well-known cascade architecture is adopted to reorganize the learned SOBoost classifier into a ’cas-cade’, by which the searching ability of iris recognition to-wards large-scale deployments is greatly enhanced. Exten-sive experiments on two challenging iris image databases demonstrate that the proposed method achieves state-of-the-art iris recognition accuracy and speed. In addi-tion, SOBoost outperforms Adaboost (Gentle-Adaboost, JS-Adaboost, etc.) in terms of both accuracy and generaliza-tion capability across different iris databases. 1