2 Citations (Scopus)

Abstract

In most existing works on shape contour matching, the shape contours are considered and matched in whole. When searching for contour snippets, however, techniques that match whole contours are not directly applicable. In particular, a relevant snippet can be anywhere on a shape contour; moreover, the relevance of shape snippet is a function of not only the shape of the snippet itself, but also of its neighborhood on the contour. In this paper, we propose an HMM based solution to shape snippet extraction. Relying on a general-purpose symbolic representation (such as SAX), we first convert the shape contour onto a representation suitable for snippet marking and extraction processes. We then show that, given a set of samples, we can train an HMM capable of detecting relevant snippets in new shape images. Next, we show that the HMM performance can be boosted significantly if the similarities between the symbolic representations are used to create new sibling training sequences from the input sequences. The experiment results show that just adding one additional sibling per input training sequence can improve the diversity of the training set sufficiently to boost the overlaps between actual and detected snippets much. While a naive application of this metadata driven training technique can increase the training costs significantly, we show that a novel metadata-driven HMM (mHMM) scheme can significantly improve the HMM-base snippet detection performance with negligible costs.

Original languageEnglish (US)
Title of host publicationCIVR 2010 - 2010 ACM International Conference on Image and Video Retrieval
Pages430-437
Number of pages8
DOIs
StatePublished - 2010
EventACM International Conference on Image and Video Retrieval, ACM-CIVR 2010 - Xi'an, China
Duration: Jul 5 2010Jul 7 2010

Other

OtherACM International Conference on Image and Video Retrieval, ACM-CIVR 2010
CountryChina
CityXi'an
Period7/5/107/7/10

Fingerprint

Hidden Markov models
Metadata
Costs
Experiments

Keywords

  • HMM
  • SAX
  • Shape snippet detection

ASJC Scopus subject areas

  • Computer Graphics and Computer-Aided Design
  • Computer Vision and Pattern Recognition
  • Human-Computer Interaction

Cite this

Wang, X., & Candan, K. (2010). Relevant shape contour snippet extraction with metadata supported hidden Markov models. In CIVR 2010 - 2010 ACM International Conference on Image and Video Retrieval (pp. 430-437) https://doi.org/10.1145/1816041.1816104

Relevant shape contour snippet extraction with metadata supported hidden Markov models. / Wang, Xinxin; Candan, Kasim.

CIVR 2010 - 2010 ACM International Conference on Image and Video Retrieval. 2010. p. 430-437.

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

Wang, X & Candan, K 2010, Relevant shape contour snippet extraction with metadata supported hidden Markov models. in CIVR 2010 - 2010 ACM International Conference on Image and Video Retrieval. pp. 430-437, ACM International Conference on Image and Video Retrieval, ACM-CIVR 2010, Xi'an, China, 7/5/10. https://doi.org/10.1145/1816041.1816104
Wang X, Candan K. Relevant shape contour snippet extraction with metadata supported hidden Markov models. In CIVR 2010 - 2010 ACM International Conference on Image and Video Retrieval. 2010. p. 430-437 https://doi.org/10.1145/1816041.1816104
Wang, Xinxin ; Candan, Kasim. / Relevant shape contour snippet extraction with metadata supported hidden Markov models. CIVR 2010 - 2010 ACM International Conference on Image and Video Retrieval. 2010. pp. 430-437
@inproceedings{6f9eee1bd90b4258804911603f62476b,
title = "Relevant shape contour snippet extraction with metadata supported hidden Markov models",
abstract = "In most existing works on shape contour matching, the shape contours are considered and matched in whole. When searching for contour snippets, however, techniques that match whole contours are not directly applicable. In particular, a relevant snippet can be anywhere on a shape contour; moreover, the relevance of shape snippet is a function of not only the shape of the snippet itself, but also of its neighborhood on the contour. In this paper, we propose an HMM based solution to shape snippet extraction. Relying on a general-purpose symbolic representation (such as SAX), we first convert the shape contour onto a representation suitable for snippet marking and extraction processes. We then show that, given a set of samples, we can train an HMM capable of detecting relevant snippets in new shape images. Next, we show that the HMM performance can be boosted significantly if the similarities between the symbolic representations are used to create new sibling training sequences from the input sequences. The experiment results show that just adding one additional sibling per input training sequence can improve the diversity of the training set sufficiently to boost the overlaps between actual and detected snippets much. While a naive application of this metadata driven training technique can increase the training costs significantly, we show that a novel metadata-driven HMM (mHMM) scheme can significantly improve the HMM-base snippet detection performance with negligible costs.",
keywords = "HMM, SAX, Shape snippet detection",
author = "Xinxin Wang and Kasim Candan",
year = "2010",
doi = "10.1145/1816041.1816104",
language = "English (US)",
isbn = "9781450301176",
pages = "430--437",
booktitle = "CIVR 2010 - 2010 ACM International Conference on Image and Video Retrieval",

}

TY - GEN

T1 - Relevant shape contour snippet extraction with metadata supported hidden Markov models

AU - Wang, Xinxin

AU - Candan, Kasim

PY - 2010

Y1 - 2010

N2 - In most existing works on shape contour matching, the shape contours are considered and matched in whole. When searching for contour snippets, however, techniques that match whole contours are not directly applicable. In particular, a relevant snippet can be anywhere on a shape contour; moreover, the relevance of shape snippet is a function of not only the shape of the snippet itself, but also of its neighborhood on the contour. In this paper, we propose an HMM based solution to shape snippet extraction. Relying on a general-purpose symbolic representation (such as SAX), we first convert the shape contour onto a representation suitable for snippet marking and extraction processes. We then show that, given a set of samples, we can train an HMM capable of detecting relevant snippets in new shape images. Next, we show that the HMM performance can be boosted significantly if the similarities between the symbolic representations are used to create new sibling training sequences from the input sequences. The experiment results show that just adding one additional sibling per input training sequence can improve the diversity of the training set sufficiently to boost the overlaps between actual and detected snippets much. While a naive application of this metadata driven training technique can increase the training costs significantly, we show that a novel metadata-driven HMM (mHMM) scheme can significantly improve the HMM-base snippet detection performance with negligible costs.

AB - In most existing works on shape contour matching, the shape contours are considered and matched in whole. When searching for contour snippets, however, techniques that match whole contours are not directly applicable. In particular, a relevant snippet can be anywhere on a shape contour; moreover, the relevance of shape snippet is a function of not only the shape of the snippet itself, but also of its neighborhood on the contour. In this paper, we propose an HMM based solution to shape snippet extraction. Relying on a general-purpose symbolic representation (such as SAX), we first convert the shape contour onto a representation suitable for snippet marking and extraction processes. We then show that, given a set of samples, we can train an HMM capable of detecting relevant snippets in new shape images. Next, we show that the HMM performance can be boosted significantly if the similarities between the symbolic representations are used to create new sibling training sequences from the input sequences. The experiment results show that just adding one additional sibling per input training sequence can improve the diversity of the training set sufficiently to boost the overlaps between actual and detected snippets much. While a naive application of this metadata driven training technique can increase the training costs significantly, we show that a novel metadata-driven HMM (mHMM) scheme can significantly improve the HMM-base snippet detection performance with negligible costs.

KW - HMM

KW - SAX

KW - Shape snippet detection

UR - http://www.scopus.com/inward/record.url?scp=77955913727&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=77955913727&partnerID=8YFLogxK

U2 - 10.1145/1816041.1816104

DO - 10.1145/1816041.1816104

M3 - Conference contribution

AN - SCOPUS:77955913727

SN - 9781450301176

SP - 430

EP - 437

BT - CIVR 2010 - 2010 ACM International Conference on Image and Video Retrieval

ER -