Presentation | 2013-05-24 Measurement of Image Instance-based Distance between Concepts Using Adaptive Visual Feature Selection Kazuaki NAKAMURA, Ayaka OTOSHI, Noboru BABAGUCHI, |
---|---|
PDF Download Page | PDF download Page Link |
Abstract(in Japanese) | (See Japanese page) |
Abstract(in English) | In recent years, measurement methods for similarity or distance between concepts have got more and more attention due to its applicability to image retrieval, annotation, clustering, and so on. Some of the methods measure the distance based on image instances of the concepts. We refer to this kind of distances as Image Instance-Based Distance, abbreviated as IIBD. Most of related works aim to compute the IIBD with a single kind of visual features extracted from each image instance. However, a single kind of features is not enough for properly measuring the IIBD, because there are diverse concepts in the world and different concepts will be well-represented by different kinds of features. In this paper, we propose to extract two or more kinds of visual features from each image instance and measure the IIBD using the multiple features. Moreover, we propose a method for adaptively selecting the visual features which are most suitable for computing the IIBD for each concept pair. In our experiments, the proposed method outperformed the methods using only a single kind of features. |
Keyword(in Japanese) | (See Japanese page) |
Keyword(in English) | distance between concepts / image instances / multiple visual features / feature selection / normalized within-class variance |
Paper # | IE2013-10,PRMU2013-3,MI2013-3 |
Date of Issue |
Conference Information | |
Committee | MI |
---|---|
Conference Date | 2013/5/17(1days) |
Place (in Japanese) | (See Japanese page) |
Place (in English) | |
Topics (in Japanese) | (See Japanese page) |
Topics (in English) | |
Chair | |
Vice Chair | |
Secretary | |
Assistant |
Paper Information | |
Registration To | Medical Imaging (MI) |
---|---|
Language | JPN |
Title (in Japanese) | (See Japanese page) |
Sub Title (in Japanese) | (See Japanese page) |
Title (in English) | Measurement of Image Instance-based Distance between Concepts Using Adaptive Visual Feature Selection |
Sub Title (in English) | |
Keyword(1) | distance between concepts |
Keyword(2) | image instances |
Keyword(3) | multiple visual features |
Keyword(4) | feature selection |
Keyword(5) | normalized within-class variance |
1st Author's Name | Kazuaki NAKAMURA |
1st Author's Affiliation | Graduate School of Engineering, Osaka University() |
2nd Author's Name | Ayaka OTOSHI |
2nd Author's Affiliation | School of Engineering, Osaka University |
3rd Author's Name | Noboru BABAGUCHI |
3rd Author's Affiliation | Graduate School of Engineering, Osaka University |
Date | 2013-05-24 |
Paper # | IE2013-10,PRMU2013-3,MI2013-3 |
Volume (vol) | vol.113 |
Number (no) | 64 |
Page | pp.pp.- |
#Pages | 6 |
Date of Issue |