New iris recognition method for noisy iris images
Highlights
► The noisy iris images degrade the accuracy of iris recognition. ► We propose a new iris recognition algorithm for noisy iris images. ► We measured the performance with the NICE.II training dataset (selected from UBIRIS.v2 database). ► Results showed the decidability value (d′) was 1.6398 (the fourth-highest rank).
Section snippets
Short introduction
Among the various biometrics methods, iris recognition achieves particularly high recognition accuracy because the iris patterns between the pupil and sclera offer rich textures with a high degree of freedom (DOF) (Daugman, 1993, Daugman, 2003, Daugman, 2004). Most iris recognition systems currently available for personal identification require near-infrared (NIR) illumination, and they require user’s cooperation to capture an iris image with good quality, which can give inconvenience to users.
Overview
Fig. 2 shows the overall procedure of the proposed method. It includes four steps: (I) iris region segmentation, (II) the 1st step classification of the “left or right eye”, (III) the 2nd step classification using “color information” of iris region, and (IV) the 3rd step classification using “textural information” of iris region. The NICE.II contest focused on iris recognition, and so the results of iris segmentation were already provided, as shown in Fig. 3b. However, the pixel positions of
Result analysis
For experiments, the NICE.II training data set selected from UBIRIS.v2 database was used, which is composed of 1000 images including 171 classes (NICE.II training dataset, 2009) (Proença et al., 2010). The image quality of this database was extremely degraded by various noise factors as shown in Fig. 1. The image resolution is 400 × 300 pixels including RGB color channels. In our experiments, the number of authentic tests was 3593 and the number of imposter tests was 495,907. The number of “left
Conclusion
We propose a new iris recognition method for the iris images degraded by noisy factors. The genuine and imposter matching are determined by pre-classification including the 1st and 2nd step classifications based on “left or right eye” and “color” information of iris. And the iris authentication is completed by comparing the iris binary code based on “texture information” of iris region. As the experimental results, the decidability value evaluated by the NICE.II contest was 1.6398, and the EER
Acknowledgments
This work was supported by the National Research Foundation of Korea (NRF) through the Biometrics Engineering Research Center (BERC) at Yonsei University [R112002105070020(2010)] and in part supported by NAP (National Agenda Project) of Korea Research Council of Fundamental Science & Technology.
References (17)
The importance of being random: statistical principles of iris recognition
Pattern Recognit.
(2003)- et al.
A new iris segmentation method for non-ideal iris images
Image Vis. Comput.
(2010) - et al.
Iris recognition based on score level fusion by using SVM
Pattern Recognit. Lett.
(2007) - Boyce, C., Ross, A., Monaco, M., Hornak, L., Li, X., 2006. Multispectral iris analysis: a preliminary study. In: Proc....
High confidence visual recognition of persons by a test of statistical independence
IEEE Trans. Pattern Anal. Mach. Intell.
(1993)How iris recognition works
IEEE Trans. Circuits Syst. Video Technol.
(2004)- et al.
Digital Image Processing
(2002) - et al.
An ocularist’s approach to human iris synthesis
IEEE Comput. Graph. Appl.
(2003)
Cited by (74)
An efficient approach for generating IRIS codes for optimally recognizing IRIS using multi objective genetic algorithm
2021, Materials Today: ProceedingsBiometric Iris Recognition System’s Software and Hardware Implementation Using LabVIEW Tool
2023, Lecture Notes in Networks and SystemsSpecular Reflection Removal Techniques for Noisy and Non-ideal Iris Images: Two New Approaches
2023, Lecture Notes in Networks and SystemsInnovative Contactless Palmprint Recognition System Based on Dual-Camera Alignment
2022, IEEE Transactions on Systems, Man, and Cybernetics: Systems