| Presentation | 2019-10-03 Recognition feature prediction from low-resolution iris images using CNN Ryo Watanabe, Keisuke Kameyama, |
|---|---|
| PDF Download Page |  PDF download Page Link |
| Abstract(in Japanese) | (See Japanese page) |
| Abstract(in English) | In iris authentication, Daugman's method employing Gabor features is widely supported. However, the method's performance suffers when there are variations in the iris observation conditions. This work aims to improve the authentication performances for cases when only low-resolution images are available upon enrollment and/or verification. In this work, we propose a method to estimate the iris features in the high-resolution (HR) images using low-resolution (LR) images of the same iris. A Convolutional Neural Network (CNN) is used to learn the relation between the local LR feature patch and the HR feature at its center. The estimated HR feature set will be used for authentication. In the experiments, improvements were not observed for the case when HR images were used for enrollment and HR featuresestimated from LR images wereused for verification. However, when HR features estimated from LRimages were used for both enrollment and verification, improvementsover the direct use of LR features were observed. |
| Keyword(in Japanese) | (See Japanese page) |
| Keyword(in English) | Iris Recognition / image feature / CNN |
| Paper # | BioX2019-55 |
| Date of Issue | 2019-09-26 (BioX) |
| Conference Information | |
| Committee | BioX |
|---|---|
| Conference Date | 2019/10/3(2days) |
| Place (in Japanese) | (See Japanese page) |
| Place (in English) | |
| Topics (in Japanese) | (See Japanese page) |
| Topics (in English) | |
| Chair | Akira Otsuka(IISEC) |
| Vice Chair | Tetsushi Ohki(Shizuoka Univ.) / Takahiro Aoki(Fujitsu Labs.) |
| Secretary | Tetsushi Ohki(Univ. of Electro-Comm.) / Takahiro Aoki(SECOM) |
| Assistant | Daishi Watabe(Saitama Inst. of Tech.) / Ryota Horie(Shibaura Inst. of Tech.) |
| Paper Information | |
| Registration To | Technical Committee on Biometrics |
|---|---|
| Language | JPN |
| Title (in Japanese) | (See Japanese page) |
| Sub Title (in Japanese) | (See Japanese page) |
| Title (in English) | Recognition feature prediction from low-resolution iris images using CNN |
| Sub Title (in English) | |
| Keyword(1) | Iris Recognition |
| Keyword(2) | image feature |
| Keyword(3) | CNN |
| 1st Author's Name | Ryo Watanabe |
| 1st Author's Affiliation | University of Tsukuba(Univ. of Tsukuba) |
| 2nd Author's Name | Keisuke Kameyama |
| 2nd Author's Affiliation | University of Tsukuba(Univ. of Tsukuba) |
| Date | 2019-10-03 |
| Paper # | BioX2019-55 |
| Volume (vol) | vol.119 |
| Number (no) | BioX-214 |
| Page | pp.pp.5-10(BioX), |
| #Pages | 6 |
| Date of Issue | 2019-09-26 (BioX) |