Multi-vector feature space based on pseudo-Euclidean space and oblique basis for similarity searches of images

Yasuo Yamane, Tadashi Hoshiai, Hiroshi Tsuda, Kaoru Katayama, Manabu Ohta, Hiroshi Ishikawa

Research output: Chapter in Book/Report/Conference proceedingConference contribution

1 Citation (Scopus)

Abstract

Investigators have tried to increase the precision of similarity searches of images by using distance functions that reflect the similarity of features. When the quadratic-form distance is used, however, dissimilar images can be judged to be similar. We therefore propose that the similarity of images be evaluated using a measure of distance in a multi-vector feature space based on pseudo-Euclidean space and an oblique basis (MVPO). In this space an image is represented by a set of vectors each of which represents each feature. And we propose a distance (called D-distance) between two sets of vectors. Roughly speaking, it is the distance between solids.Another representative distance used in similarity searches is the Earth Mover's Distance (EMD). It can be formalized using MVPO, and that explains well why EMD outperforms quad-ratic-form distance. The main difference between EMD and D-distance is that EMD is based on partial matching and D-distance is based on total matching.We also discuss performance issues of MPVO and D-distance to address practical use of them.

Original languageEnglish
Title of host publicationACM International Conference Proceeding Series
Pages27-34
Number of pages8
Volume66
DOIs
Publication statusPublished - 2004
Externally publishedYes
Event1st International Workshop on Computer Vision Meets Databases, CVDB 2004 - Paris, France
Duration: Jun 13 2004Jun 13 2004

Other

Other1st International Workshop on Computer Vision Meets Databases, CVDB 2004
CountryFrance
CityParis
Period6/13/046/13/04

Fingerprint

Earth (planet)

ASJC Scopus subject areas

  • Human-Computer Interaction
  • Computer Networks and Communications
  • Computer Vision and Pattern Recognition
  • Software

Cite this

Yamane, Y., Hoshiai, T., Tsuda, H., Katayama, K., Ohta, M., & Ishikawa, H. (2004). Multi-vector feature space based on pseudo-Euclidean space and oblique basis for similarity searches of images. In ACM International Conference Proceeding Series (Vol. 66, pp. 27-34) https://doi.org/10.1145/1039470.1039479

Multi-vector feature space based on pseudo-Euclidean space and oblique basis for similarity searches of images. / Yamane, Yasuo; Hoshiai, Tadashi; Tsuda, Hiroshi; Katayama, Kaoru; Ohta, Manabu; Ishikawa, Hiroshi.

ACM International Conference Proceeding Series. Vol. 66 2004. p. 27-34.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Yamane, Y, Hoshiai, T, Tsuda, H, Katayama, K, Ohta, M & Ishikawa, H 2004, Multi-vector feature space based on pseudo-Euclidean space and oblique basis for similarity searches of images. in ACM International Conference Proceeding Series. vol. 66, pp. 27-34, 1st International Workshop on Computer Vision Meets Databases, CVDB 2004, Paris, France, 6/13/04. https://doi.org/10.1145/1039470.1039479
Yamane Y, Hoshiai T, Tsuda H, Katayama K, Ohta M, Ishikawa H. Multi-vector feature space based on pseudo-Euclidean space and oblique basis for similarity searches of images. In ACM International Conference Proceeding Series. Vol. 66. 2004. p. 27-34 https://doi.org/10.1145/1039470.1039479
Yamane, Yasuo ; Hoshiai, Tadashi ; Tsuda, Hiroshi ; Katayama, Kaoru ; Ohta, Manabu ; Ishikawa, Hiroshi. / Multi-vector feature space based on pseudo-Euclidean space and oblique basis for similarity searches of images. ACM International Conference Proceeding Series. Vol. 66 2004. pp. 27-34
@inproceedings{3b3db350d7ef40eab10ddc68a039e6fa,
title = "Multi-vector feature space based on pseudo-Euclidean space and oblique basis for similarity searches of images",
abstract = "Investigators have tried to increase the precision of similarity searches of images by using distance functions that reflect the similarity of features. When the quadratic-form distance is used, however, dissimilar images can be judged to be similar. We therefore propose that the similarity of images be evaluated using a measure of distance in a multi-vector feature space based on pseudo-Euclidean space and an oblique basis (MVPO). In this space an image is represented by a set of vectors each of which represents each feature. And we propose a distance (called D-distance) between two sets of vectors. Roughly speaking, it is the distance between solids.Another representative distance used in similarity searches is the Earth Mover's Distance (EMD). It can be formalized using MVPO, and that explains well why EMD outperforms quad-ratic-form distance. The main difference between EMD and D-distance is that EMD is based on partial matching and D-distance is based on total matching.We also discuss performance issues of MPVO and D-distance to address practical use of them.",
author = "Yasuo Yamane and Tadashi Hoshiai and Hiroshi Tsuda and Kaoru Katayama and Manabu Ohta and Hiroshi Ishikawa",
year = "2004",
doi = "10.1145/1039470.1039479",
language = "English",
isbn = "1581139179",
volume = "66",
pages = "27--34",
booktitle = "ACM International Conference Proceeding Series",

}

TY - GEN

T1 - Multi-vector feature space based on pseudo-Euclidean space and oblique basis for similarity searches of images

AU - Yamane, Yasuo

AU - Hoshiai, Tadashi

AU - Tsuda, Hiroshi

AU - Katayama, Kaoru

AU - Ohta, Manabu

AU - Ishikawa, Hiroshi

PY - 2004

Y1 - 2004

N2 - Investigators have tried to increase the precision of similarity searches of images by using distance functions that reflect the similarity of features. When the quadratic-form distance is used, however, dissimilar images can be judged to be similar. We therefore propose that the similarity of images be evaluated using a measure of distance in a multi-vector feature space based on pseudo-Euclidean space and an oblique basis (MVPO). In this space an image is represented by a set of vectors each of which represents each feature. And we propose a distance (called D-distance) between two sets of vectors. Roughly speaking, it is the distance between solids.Another representative distance used in similarity searches is the Earth Mover's Distance (EMD). It can be formalized using MVPO, and that explains well why EMD outperforms quad-ratic-form distance. The main difference between EMD and D-distance is that EMD is based on partial matching and D-distance is based on total matching.We also discuss performance issues of MPVO and D-distance to address practical use of them.

AB - Investigators have tried to increase the precision of similarity searches of images by using distance functions that reflect the similarity of features. When the quadratic-form distance is used, however, dissimilar images can be judged to be similar. We therefore propose that the similarity of images be evaluated using a measure of distance in a multi-vector feature space based on pseudo-Euclidean space and an oblique basis (MVPO). In this space an image is represented by a set of vectors each of which represents each feature. And we propose a distance (called D-distance) between two sets of vectors. Roughly speaking, it is the distance between solids.Another representative distance used in similarity searches is the Earth Mover's Distance (EMD). It can be formalized using MVPO, and that explains well why EMD outperforms quad-ratic-form distance. The main difference between EMD and D-distance is that EMD is based on partial matching and D-distance is based on total matching.We also discuss performance issues of MPVO and D-distance to address practical use of them.

UR - http://www.scopus.com/inward/record.url?scp=77952978645&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=77952978645&partnerID=8YFLogxK

U2 - 10.1145/1039470.1039479

DO - 10.1145/1039470.1039479

M3 - Conference contribution

SN - 1581139179

SN - 9781581139174

VL - 66

SP - 27

EP - 34

BT - ACM International Conference Proceeding Series

ER -