Focus assessment measure and 2DLDA features for iris recognition(Biometrics1)

概要

論文の詳細を見る
In this paper, we propose a robust and real-time applicable iris recognition system, which has focus assessment measure and uses 2DLDA(2 Dimensional Linear Discriminant Analysis) method for feature extraction. The acquired image captured by iris camera is evaluated by focus assessment measure based on wavelet transform and SVM(Support Vector Machine). The iris area is extracted by the Canny Edge detector and bisection based Hough transform. For comparing the recognition rate, we used the YS020611 and CASIA database and tests are performed on original and transformed images by the wavelet to 1-level and 2-level. In addition, we compare the recognition rates of other feature extraction methods such as PCA, LDA, PCA+LDA, 2DPCA and Garbor filter. The smallest EER(Equal Error rate) is 0.0001% when we use the 2DLDA with iris image transformed by the wavelet to 1-level.
社団法人電子情報通信学会の論文
2006-11-17