| Presentation | 2019-05-23 Generation of privacy-preserving images holding positional information for HOG feature extraction Masaki Kitayama, Hitoshi Kiya, |
|---|---|
| PDF Download Page |  PDF download Page Link |
| Abstract(in Japanese) | (See Japanese page) |
| Abstract(in English) | In this paper, we propose a generation method of images which have no visual information but hold the gradient direction information of the original images. Then we propose an extraction method of Histogram-of-Oriented-Gradients (HOG) features from the images (privacy-preserved images), and apply the features to machine learning algorithms. The proposed generation method of privacy-preserved images is an irreversible method without any encryption keys. Thereby, unlike conventional reversible methods to protect visual information by using encryption keys, the proposed method does not need to consider secure management and transfer of the encryption keys. Moreover, the proposed method solves the huge culculation cost problem that conventional methods based on homomorphic encryption have. In this paper, in addition to a novel generation method of privacy-preserved images, an effective method of extracting HOG features from the images is also considered. In an experiment, a face classification task is carried out under the use of support vector machine with the HOG features to confirm the effectiveness of the proposed method. |
| Keyword(in Japanese) | (See Japanese page) |
| Keyword(in English) | encryption / privacy-preserving / machine learning / HOG features |
| Paper # | IT2019-1,EMM2019-1 |
| Date of Issue | 2019-05-16 (IT, EMM) |
| Conference Information | |
| Committee | EMM / IT |
|---|---|
| Conference Date | 2019/5/23(2days) |
| Place (in Japanese) | (See Japanese page) |
| Place (in English) | Asahikawa International Conference Hall |
| Topics (in Japanese) | (See Japanese page) |
| Topics (in English) | Information Security, Information Theory, Information Hiding, etc. |
| Chair | Keiichi Iwamura(TUC) / Jun Muramatsu(NTT) |
| Vice Chair | Minoru Kuribayashi(Okayama Univ.) / Tetsuya Kojima(NIT,Tokyo College) / Tadashi Wadayama(Nagoya Inst. of Tech.) |
| Secretary | Minoru Kuribayashi(NIT, Tokyo) / Tetsuya Kojima(Tyukyo Univ.) / Tadashi Wadayama(Nagano Pref Inst. of Tech.) |
| Assistant | Hiroko Akiyama(NIT, Nagano College) / キタヒロ カネダ(CANON) / Takahiro Yoshida(Yokohama College of Commerce) |
| Paper Information | |
| Registration To | Technical Committee on Enriched MultiMedia / Technical Committee on Information Theory |
|---|---|
| Language | JPN |
| Title (in Japanese) | (See Japanese page) |
| Sub Title (in Japanese) | (See Japanese page) |
| Title (in English) | Generation of privacy-preserving images holding positional information for HOG feature extraction |
| Sub Title (in English) | |
| Keyword(1) | encryption |
| Keyword(2) | privacy-preserving |
| Keyword(3) | machine learning |
| Keyword(4) | HOG features |
| 1st Author's Name | Masaki Kitayama |
| 1st Author's Affiliation | Tokyo Metropolitan University(Tokyo Metro. Univ.) |
| 2nd Author's Name | Hitoshi Kiya |
| 2nd Author's Affiliation | Tokyo Metropolitan University(Tokyo Metro. Univ.) |
| Date | 2019-05-23 |
| Paper # | IT2019-1,EMM2019-1 |
| Volume (vol) | vol.119 |
| Number (no) | IT-47,EMM-48 |
| Page | pp.pp.1-6(IT), pp.1-6(EMM), |
| #Pages | 6 |
| Date of Issue | 2019-05-16 (IT, EMM) |