Paper Abstract and Keywords |
Presentation |
2010-12-19 13:30
A Hierarchal Features Realization Method by Self-Organizing Map Shuta Saito, Masahiro Ariizumi, Hakaru Tamukoh, Masatoshi Sekine (Tokyo Univ. of Agr and Tech) MBE2010-72 NC2010-83 |
Abstract |
(in Japanese) |
(See Japanese page) |
(in English) |
In this paper, we propose a tree structured SOM to learn a base vector to express the feature of the image hierarchically with Wavelet transformation, clustering by Self-organizing map and Gram-Schmidt orthonormalization. With the making of the base vector, this system search for a feature region step by step and extract an existing local feature and an existing global feature every each resolution level. We was able to extract part neighborhood such as eyes and the mouth in the experiment for the face image. In addition, we generated a template image from the feature of an image built in tree structure type and performed a face image recognition experiment by the template matching. The utility of the feature subspace base vector of the tree structure type was confirmed. |
Keyword |
(in Japanese) |
(See Japanese page) |
(in English) |
face image recognition / wavelet transform / Self-Organizing Map / feature region / template matching / / / |
Reference Info. |
IEICE Tech. Rep., vol. 110, no. 355, NC2010-83, pp. 97-102, Dec. 2010. |
Paper # |
NC2010-83 |
Date of Issue |
2010-12-12 (MBE, NC) |
ISSN |
Print edition: ISSN 0913-5685 Online edition: ISSN 2432-6380 |
Copyright and reproduction |
All rights are reserved and no part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopy, recording, or any information storage and retrieval system, without permission in writing from the publisher. Notwithstanding, instructors are permitted to photocopy isolated articles for noncommercial classroom use without fee. (License No.: 10GA0019/12GB0052/13GB0056/17GB0034/18GB0034) |
Download PDF |
MBE2010-72 NC2010-83 |
|