Presentation 2019-05-24
Hierarchical Personal Identification in Short and Long Distance using Blink Motion Features
Mariko Nakano, Ritsuko Tokunaga, Yoko Uchida, Daisuke Sugimura,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) We propose a method for person identification using blink motion features. Previous methods are difficult to achieve a person identification with accuracy in a long distance from the camera. Furthermore, the classification accuracy decreases as the number of subjects to be identified increases. To address these problems, our method performs hierarchical classifications in short and long distances. We first perform a group-level identification using roughly-discriminative features in blink motions taken in a long distance. We then identify a person in the group using fine blink motion features taken in a short distance. Our hierarchical approach improves the performance of the person identification because the number ofcandidates can be reduced by our group-level classification. Experiments using blink motion videos taken by ourselves demonstrate the effectiveness of the proposed method. Furthermore, we investigate influences of video spatial resolutions and eye fixations for the performance of the person identification.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Person identification / Blink motion / Hierarchical classification
Paper # SIP2019-7,IE2019-7,MI2019-7
Date of Issue 2019-05-16 (SIP, IE, MI)

Conference Information
Committee SIP / MI / IE
Conference Date 2019/5/23(2days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair Shogo Muramatsu(Niigata Univ.) / Kensaku Mori(Nagoya Univ.) / Takayuki Hamamoto(Tokyo Univ. of Science)
Vice Chair Naoyuki Aikawa(TUS) / Kazunori Hayashi(Osaka City Univ) / Yoshiki Kawata(Tokushima Univ.) / Yuichi Kimura(Kinki Univ.) / Hideaki Kimata(NTT) / Kazuya Kodama(NII)
Secretary Naoyuki Aikawa(Takushoku Univ.) / Kazunori Hayashi(Hiroshima Univ.) / Yoshiki Kawata(Aichi Inst. of Tech.) / Yuichi Kimura(Nagoya Inst. of Tech.) / Hideaki Kimata(KDDI Research) / Kazuya Kodama(Nagoya Univ.)
Assistant Katsumi Konishi(Hosei Univ.) / Ryo Haraguchi(Univ. of Hyogo) / Yasushi Hirano(Yamaguchi Univ.) / Kazuya Hayase(NTT) / Yasutaka Matsuo(NHK)

Paper Information
Registration To Technical Committee on Signal Processing / Technical Committee on Medical Imaging / Technical Committee on Image Engineering
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Hierarchical Personal Identification in Short and Long Distance using Blink Motion Features
Sub Title (in English)
Keyword(1) Person identification
Keyword(2) Blink motion
Keyword(3) Hierarchical classification
1st Author's Name Mariko Nakano
1st Author's Affiliation Tsuda University(Tsuda Univ.)
2nd Author's Name Ritsuko Tokunaga
2nd Author's Affiliation Tsuda University(Tsuda Univ.)
3rd Author's Name Yoko Uchida
3rd Author's Affiliation Tsuda University(Tsuda Univ.)
4th Author's Name Daisuke Sugimura
4th Author's Affiliation Tsuda University(Tsuda Univ.)
Date 2019-05-24
Paper # SIP2019-7,IE2019-7,MI2019-7
Volume (vol) vol.119
Number (no) SIP-49,IE-50,MI-51
Page pp.pp.29-32(SIP), pp.29-32(IE), pp.29-32(MI),
#Pages 4
Date of Issue 2019-05-16 (SIP, IE, MI)