EREL: Extremal Regions of Extremum Levels

Mehdi Faraji, Jamshid Shanbehzadeh, Kamal Nasrollahi, Thomas B. Moeslund

Research output: Contribution to book/anthology/report/conference proceedingArticle in proceedingResearchpeer-review

12 Citations (Scopus)
386 Downloads (Pure)

Abstract

Maximally Stable Extremal Regions (MSER) is a novel region detector. It has been employed in many applications in order to extract affine covariant regions. Although MSER results in regions with almost high repeatability, it is heavily dependent on the union-find approach which is a fairly complicated algorithm, and should be completed sequentially. Furthermore, it detects regions with low repeatability under the blur transformations. The reason for the latter shortcoming is the absence of boundaries information in stability criterion. To tackle these problems we propose to employ prior information about boundaries of regions, which results in a novel region detector algorithm that not only outperforms MSER, but avoids the MSER's rather complicated steps of union-finding. To achieve that, we introduce Maxima of Gradient Magnitudes (MGMs) and use them to find handful of Extremum Levels (ELs). The chosen ELs are then scanned to detect their Extremal Regions (ER). The proposed algorithm which is called Extremal Regions of Extremum Levels (EREL) has been tested on the public benchmark dataset of Mikolajczyk. Our experimental evaluations illustrate that, in many cases EREL achieves higher repeatability scores than MSER even for very low overlap errors.
Original languageEnglish
Title of host publicationIEEE International Conference on Image Processing (ICIP), 2015
PublisherIEEE Signal Processing Society
Publication date2015
Pages681-685
ISBN (Print)978-1-4799-8339-1
DOIs
Publication statusPublished - 2015
EventIEEE International Conference on Image Processing - Québec City, Canada
Duration: 27 Sep 201530 Sep 2015

Conference

ConferenceIEEE International Conference on Image Processing
CountryCanada
City Québec City
Period27/09/201530/09/2015

Fingerprint

Detectors
Stability criteria

Keywords

  • Maxima of Gradient Magnitude (MGM),
  • Maximally Stable Extremal Region (MSER)
  • Extremal Regions of Extremum Levels (EREL),
  • Feature Detection

Cite this

Faraji, M., Shanbehzadeh, J., Nasrollahi, K., & Moeslund, T. B. (2015). EREL: Extremal Regions of Extremum Levels. In IEEE International Conference on Image Processing (ICIP), 2015 (pp. 681-685). IEEE Signal Processing Society. https://doi.org/10.1109/ICIP.2015.7350885
Faraji, Mehdi ; Shanbehzadeh, Jamshid ; Nasrollahi, Kamal ; Moeslund, Thomas B. / EREL : Extremal Regions of Extremum Levels. IEEE International Conference on Image Processing (ICIP), 2015. IEEE Signal Processing Society, 2015. pp. 681-685
@inproceedings{d7a340d610864c61bee6f88306a7bd97,
title = "EREL: Extremal Regions of Extremum Levels",
abstract = "Maximally Stable Extremal Regions (MSER) is a novel region detector. It has been employed in many applications in order to extract affine covariant regions. Although MSER results in regions with almost high repeatability, it is heavily dependent on the union-find approach which is a fairly complicated algorithm, and should be completed sequentially. Furthermore, it detects regions with low repeatability under the blur transformations. The reason for the latter shortcoming is the absence of boundaries information in stability criterion. To tackle these problems we propose to employ prior information about boundaries of regions, which results in a novel region detector algorithm that not only outperforms MSER, but avoids the MSER's rather complicated steps of union-finding. To achieve that, we introduce Maxima of Gradient Magnitudes (MGMs) and use them to find handful of Extremum Levels (ELs). The chosen ELs are then scanned to detect their Extremal Regions (ER). The proposed algorithm which is called Extremal Regions of Extremum Levels (EREL) has been tested on the public benchmark dataset of Mikolajczyk. Our experimental evaluations illustrate that, in many cases EREL achieves higher repeatability scores than MSER even for very low overlap errors.",
keywords = "Maxima of Gradient Magnitude (MGM),, Maximally Stable Extremal Region (MSER), Extremal Regions of Extremum Levels (EREL),, Feature Detection",
author = "Mehdi Faraji and Jamshid Shanbehzadeh and Kamal Nasrollahi and Moeslund, {Thomas B.}",
year = "2015",
doi = "10.1109/ICIP.2015.7350885",
language = "English",
isbn = "978-1-4799-8339-1",
pages = "681--685",
booktitle = "IEEE International Conference on Image Processing (ICIP), 2015",
publisher = "IEEE Signal Processing Society",
address = "United States",

}

Faraji, M, Shanbehzadeh, J, Nasrollahi, K & Moeslund, TB 2015, EREL: Extremal Regions of Extremum Levels. in IEEE International Conference on Image Processing (ICIP), 2015. IEEE Signal Processing Society, pp. 681-685, IEEE International Conference on Image Processing, Québec City, Canada, 27/09/2015. https://doi.org/10.1109/ICIP.2015.7350885

EREL : Extremal Regions of Extremum Levels. / Faraji, Mehdi; Shanbehzadeh, Jamshid; Nasrollahi, Kamal; Moeslund, Thomas B.

IEEE International Conference on Image Processing (ICIP), 2015. IEEE Signal Processing Society, 2015. p. 681-685.

Research output: Contribution to book/anthology/report/conference proceedingArticle in proceedingResearchpeer-review

TY - GEN

T1 - EREL

T2 - Extremal Regions of Extremum Levels

AU - Faraji, Mehdi

AU - Shanbehzadeh, Jamshid

AU - Nasrollahi, Kamal

AU - Moeslund, Thomas B.

PY - 2015

Y1 - 2015

N2 - Maximally Stable Extremal Regions (MSER) is a novel region detector. It has been employed in many applications in order to extract affine covariant regions. Although MSER results in regions with almost high repeatability, it is heavily dependent on the union-find approach which is a fairly complicated algorithm, and should be completed sequentially. Furthermore, it detects regions with low repeatability under the blur transformations. The reason for the latter shortcoming is the absence of boundaries information in stability criterion. To tackle these problems we propose to employ prior information about boundaries of regions, which results in a novel region detector algorithm that not only outperforms MSER, but avoids the MSER's rather complicated steps of union-finding. To achieve that, we introduce Maxima of Gradient Magnitudes (MGMs) and use them to find handful of Extremum Levels (ELs). The chosen ELs are then scanned to detect their Extremal Regions (ER). The proposed algorithm which is called Extremal Regions of Extremum Levels (EREL) has been tested on the public benchmark dataset of Mikolajczyk. Our experimental evaluations illustrate that, in many cases EREL achieves higher repeatability scores than MSER even for very low overlap errors.

AB - Maximally Stable Extremal Regions (MSER) is a novel region detector. It has been employed in many applications in order to extract affine covariant regions. Although MSER results in regions with almost high repeatability, it is heavily dependent on the union-find approach which is a fairly complicated algorithm, and should be completed sequentially. Furthermore, it detects regions with low repeatability under the blur transformations. The reason for the latter shortcoming is the absence of boundaries information in stability criterion. To tackle these problems we propose to employ prior information about boundaries of regions, which results in a novel region detector algorithm that not only outperforms MSER, but avoids the MSER's rather complicated steps of union-finding. To achieve that, we introduce Maxima of Gradient Magnitudes (MGMs) and use them to find handful of Extremum Levels (ELs). The chosen ELs are then scanned to detect their Extremal Regions (ER). The proposed algorithm which is called Extremal Regions of Extremum Levels (EREL) has been tested on the public benchmark dataset of Mikolajczyk. Our experimental evaluations illustrate that, in many cases EREL achieves higher repeatability scores than MSER even for very low overlap errors.

KW - Maxima of Gradient Magnitude (MGM),

KW - Maximally Stable Extremal Region (MSER)

KW - Extremal Regions of Extremum Levels (EREL),

KW - Feature Detection

U2 - 10.1109/ICIP.2015.7350885

DO - 10.1109/ICIP.2015.7350885

M3 - Article in proceeding

SN - 978-1-4799-8339-1

SP - 681

EP - 685

BT - IEEE International Conference on Image Processing (ICIP), 2015

PB - IEEE Signal Processing Society

ER -

Faraji M, Shanbehzadeh J, Nasrollahi K, Moeslund TB. EREL: Extremal Regions of Extremum Levels. In IEEE International Conference on Image Processing (ICIP), 2015. IEEE Signal Processing Society. 2015. p. 681-685 https://doi.org/10.1109/ICIP.2015.7350885