Scene classification using spatial relationship between local posterior probabilities

Tetsu Matsukawa, Takio Kurita

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

3 Citations (Scopus)

Abstract

This paper presents scene classification methods using spatial relationship between local posterior probabilities of each category. Recently, the authors proposed the probability higher-order local autocorrelations (PHLAC) feature. This method uses autocorrelations of local posterior probabilities to capture spatial distributions of local posterior probabilities of a category. Although PHLAC achieves good recognition accuracies for scene classification, we can improve the performance further by using crosscorrelation between categories. We extend PHLAC features to crosscorrelations of posterior probabilities of other categories. Also, we introduce the subtraction operator for describing another spatial relationship of local posterior probabilities, and present vertical/horizontal mask patterns for the spatial layout of auto/ crosscorrelations. Since the combination of category index is large, we compress the proposed features by two-dimensional principal component analysis. We confirmed the effectiveness of the proposed methods using Scene-15 dataset, and our method exhibited competitive performances to recent methods without using spatial grid informations and even using linear classifiers.

Original languageEnglish
Title of host publicationVISAPP 2010 - Proceedings of the International Conference on Computer Vision Theory and Applications
Pages325-332
Number of pages8
Publication statusPublished - Sep 10 2010
Event5th International Conference on Computer Vision Theory and Applications, VISAPP 2010 - Angers, France
Duration: May 17 2010May 21 2010

Publication series

NameVISAPP 2010 - Proceedings of the International Conference on Computer Vision Theory and Applications
Volume2

Other

Other5th International Conference on Computer Vision Theory and Applications, VISAPP 2010
CountryFrance
CityAngers
Period5/17/105/21/10

Fingerprint

Autocorrelation
Principal component analysis
Spatial distribution
Masks
Classifiers

All Science Journal Classification (ASJC) codes

  • Computational Theory and Mathematics
  • Computer Science Applications
  • Computer Vision and Pattern Recognition

Cite this

Matsukawa, T., & Kurita, T. (2010). Scene classification using spatial relationship between local posterior probabilities. In VISAPP 2010 - Proceedings of the International Conference on Computer Vision Theory and Applications (pp. 325-332). (VISAPP 2010 - Proceedings of the International Conference on Computer Vision Theory and Applications; Vol. 2).

Scene classification using spatial relationship between local posterior probabilities. / Matsukawa, Tetsu; Kurita, Takio.

VISAPP 2010 - Proceedings of the International Conference on Computer Vision Theory and Applications. 2010. p. 325-332 (VISAPP 2010 - Proceedings of the International Conference on Computer Vision Theory and Applications; Vol. 2).

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

Matsukawa, T & Kurita, T 2010, Scene classification using spatial relationship between local posterior probabilities. in VISAPP 2010 - Proceedings of the International Conference on Computer Vision Theory and Applications. VISAPP 2010 - Proceedings of the International Conference on Computer Vision Theory and Applications, vol. 2, pp. 325-332, 5th International Conference on Computer Vision Theory and Applications, VISAPP 2010, Angers, France, 5/17/10.
Matsukawa T, Kurita T. Scene classification using spatial relationship between local posterior probabilities. In VISAPP 2010 - Proceedings of the International Conference on Computer Vision Theory and Applications. 2010. p. 325-332. (VISAPP 2010 - Proceedings of the International Conference on Computer Vision Theory and Applications).
Matsukawa, Tetsu ; Kurita, Takio. / Scene classification using spatial relationship between local posterior probabilities. VISAPP 2010 - Proceedings of the International Conference on Computer Vision Theory and Applications. 2010. pp. 325-332 (VISAPP 2010 - Proceedings of the International Conference on Computer Vision Theory and Applications).
@inproceedings{01fdc19ce2b24834b4d6b3a3279927c9,
title = "Scene classification using spatial relationship between local posterior probabilities",
abstract = "This paper presents scene classification methods using spatial relationship between local posterior probabilities of each category. Recently, the authors proposed the probability higher-order local autocorrelations (PHLAC) feature. This method uses autocorrelations of local posterior probabilities to capture spatial distributions of local posterior probabilities of a category. Although PHLAC achieves good recognition accuracies for scene classification, we can improve the performance further by using crosscorrelation between categories. We extend PHLAC features to crosscorrelations of posterior probabilities of other categories. Also, we introduce the subtraction operator for describing another spatial relationship of local posterior probabilities, and present vertical/horizontal mask patterns for the spatial layout of auto/ crosscorrelations. Since the combination of category index is large, we compress the proposed features by two-dimensional principal component analysis. We confirmed the effectiveness of the proposed methods using Scene-15 dataset, and our method exhibited competitive performances to recent methods without using spatial grid informations and even using linear classifiers.",
author = "Tetsu Matsukawa and Takio Kurita",
year = "2010",
month = "9",
day = "10",
language = "English",
isbn = "9789896740283",
series = "VISAPP 2010 - Proceedings of the International Conference on Computer Vision Theory and Applications",
pages = "325--332",
booktitle = "VISAPP 2010 - Proceedings of the International Conference on Computer Vision Theory and Applications",

}

TY - GEN

T1 - Scene classification using spatial relationship between local posterior probabilities

AU - Matsukawa, Tetsu

AU - Kurita, Takio

PY - 2010/9/10

Y1 - 2010/9/10

N2 - This paper presents scene classification methods using spatial relationship between local posterior probabilities of each category. Recently, the authors proposed the probability higher-order local autocorrelations (PHLAC) feature. This method uses autocorrelations of local posterior probabilities to capture spatial distributions of local posterior probabilities of a category. Although PHLAC achieves good recognition accuracies for scene classification, we can improve the performance further by using crosscorrelation between categories. We extend PHLAC features to crosscorrelations of posterior probabilities of other categories. Also, we introduce the subtraction operator for describing another spatial relationship of local posterior probabilities, and present vertical/horizontal mask patterns for the spatial layout of auto/ crosscorrelations. Since the combination of category index is large, we compress the proposed features by two-dimensional principal component analysis. We confirmed the effectiveness of the proposed methods using Scene-15 dataset, and our method exhibited competitive performances to recent methods without using spatial grid informations and even using linear classifiers.

AB - This paper presents scene classification methods using spatial relationship between local posterior probabilities of each category. Recently, the authors proposed the probability higher-order local autocorrelations (PHLAC) feature. This method uses autocorrelations of local posterior probabilities to capture spatial distributions of local posterior probabilities of a category. Although PHLAC achieves good recognition accuracies for scene classification, we can improve the performance further by using crosscorrelation between categories. We extend PHLAC features to crosscorrelations of posterior probabilities of other categories. Also, we introduce the subtraction operator for describing another spatial relationship of local posterior probabilities, and present vertical/horizontal mask patterns for the spatial layout of auto/ crosscorrelations. Since the combination of category index is large, we compress the proposed features by two-dimensional principal component analysis. We confirmed the effectiveness of the proposed methods using Scene-15 dataset, and our method exhibited competitive performances to recent methods without using spatial grid informations and even using linear classifiers.

UR - http://www.scopus.com/inward/record.url?scp=77956302021&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=77956302021&partnerID=8YFLogxK

M3 - Conference contribution

SN - 9789896740283

T3 - VISAPP 2010 - Proceedings of the International Conference on Computer Vision Theory and Applications

SP - 325

EP - 332

BT - VISAPP 2010 - Proceedings of the International Conference on Computer Vision Theory and Applications

ER -