Coarse Iris Classification Using Histogram Statistics Features
Rongnian Tang, Shaojie Weng, Yong Wang · Guangdian gongcheng · 2011
Aimed to improving the performance of iris matching, a new method for the coarse classification of iris images using Histogram statistics features is proposed. First, the iris image is segmented into eight blocks. Then, we calculate the Histogram ratio of these image blocks and take them as the class feature of iris image type. Finally, the iris images are classified into five categories in accordance with the similarity of blocks. The proposed method has been tested and evaluated on 500 iris cases from CASIA iris database. The experiment results shows that the proposed method classifies iris types with an accuracy of 98.2%. Compared with the initial Daugman iris matching method, a modified matching method, the proposed method with Daugman method, has lower equal error rate and decreases 29.4% matching time.