Benchmarking Segmentation Results Using a Markov Model and a Bayes Information Criterion

Fionn Murtagh; Xiaoyu Qiao; Danny Crookes; Paul Walsh; P. A.Muhammed Basheer; Adrian Long

doi:10.1117/12.467441

Benchmarking Segmentation Results Using a Markov Model and a Bayes Information Criterion

Fionn Murtagh, Xiaoyu Qiao, Danny Crookes, Paul Walsh, P. A.Muhammed Basheer, Adrian Long

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

3 Citations (Scopus)

Abstract

Features are derived from wavelet transforms of images containing a mixture of textures. In each case, the texture mixture is segmented, based on a 10-dimensional feature vector associated with every pixel. We show that the quality of the resulting segmentations can be characterized using the Potts or Ising spatial homogeneity parameter. This measure is defined from the segmentation labels. In order to have a better measure which takes into account both the segmentation labels and the input data, we determine the likelihood of the observed data given the model, which in turn is directly related to the Bayes information criterion, BIC. Finally we discuss how BIC is used as an approximation in model assessment using a Bayes factor.

Original language	English
Title of host publication	Opto-Ireland 2002
Subtitle of host publication	Optical Metrology, Imaging, and Machine Vision
Editors	Andrew Shearer, Fionn D. Murtagh, James Mahon, Paul F. Whelan
Publisher	SPIE
Pages	248-254
Number of pages	7
Volume	4877
ISBN (Print)	0819446580, 9780819446589
DOIs	https://doi.org/10.1117/12.467441
Publication status	Published - 19 Mar 2003
Externally published	Yes
Event	Opto-Ireland 2002: Optical metrology, Imaging, and Machine Vision - Galway, Ireland Duration: 5 Sep 2002 → 6 Sep 2002 https://www.researchgate.net/publication/260899755_Opto-Ireland_2002_Optical_Metrology_Imaging_and_Machine_Vision

Publication series

Name	Proceedings of SPIE - The International Society for Optical Engineering
Publisher	SPIE
Volume	4877
ISSN (Print)	0277-786X

Conference

Conference	Opto-Ireland 2002: Optical metrology, Imaging, and Machine Vision
Country/Territory	Ireland
City	Galway
Period	5/09/02 → 6/09/02
Internet address	https://www.researchgate.net/publication/260899755_Opto-Ireland_2002_Optical_Metrology_Imaging_and_Machine_Vision

Access to Document

10.1117/12.467441Licence: Unspecified

Cite this

Murtagh, F., Qiao, X., Crookes, D., Walsh, P., Basheer, P. A. M., & Long, A. (2003). Benchmarking Segmentation Results Using a Markov Model and a Bayes Information Criterion. In A. Shearer, F. D. Murtagh, J. Mahon, & P. F. Whelan (Eds.), Opto-Ireland 2002: Optical Metrology, Imaging, and Machine Vision (Vol. 4877, pp. 248-254). (Proceedings of SPIE - The International Society for Optical Engineering; Vol. 4877). SPIE. https://doi.org/10.1117/12.467441

Murtagh, Fionn ; Qiao, Xiaoyu ; Crookes, Danny et al. / Benchmarking Segmentation Results Using a Markov Model and a Bayes Information Criterion. Opto-Ireland 2002: Optical Metrology, Imaging, and Machine Vision. editor / Andrew Shearer ; Fionn D. Murtagh ; James Mahon ; Paul F. Whelan. Vol. 4877 SPIE, 2003. pp. 248-254 (Proceedings of SPIE - The International Society for Optical Engineering).

@inproceedings{0dca4e1f70cd4340ade17063cf9b3a1a,

title = "Benchmarking Segmentation Results Using a Markov Model and a Bayes Information Criterion",

abstract = "Features are derived from wavelet transforms of images containing a mixture of textures. In each case, the texture mixture is segmented, based on a 10-dimensional feature vector associated with every pixel. We show that the quality of the resulting segmentations can be characterized using the Potts or Ising spatial homogeneity parameter. This measure is defined from the segmentation labels. In order to have a better measure which takes into account both the segmentation labels and the input data, we determine the likelihood of the observed data given the model, which in turn is directly related to the Bayes information criterion, BIC. Finally we discuss how BIC is used as an approximation in model assessment using a Bayes factor.",

keywords = "Image segmentation, Data modeling, Expectation maximization algorithms, Wavelet transforms, Performance modeling, Wavelets, Algorithm development, Visualization, Civil engineering, Computer science",

author = "Fionn Murtagh and Xiaoyu Qiao and Danny Crookes and Paul Walsh and Basheer, {P. A.Muhammed} and Adrian Long",

year = "2003",

month = mar,

day = "19",

doi = "10.1117/12.467441",

language = "English",

isbn = "0819446580",

volume = "4877",

series = "Proceedings of SPIE - The International Society for Optical Engineering",

publisher = "SPIE",

pages = "248--254",

editor = "Andrew Shearer and Murtagh, {Fionn D.} and James Mahon and Whelan, {Paul F.}",

booktitle = "Opto-Ireland 2002",

address = "United States",

note = "Opto-Ireland 2002: Optical metrology, Imaging, and Machine Vision ; Conference date: 05-09-2002 Through 06-09-2002",

url = "https://www.researchgate.net/publication/260899755_Opto-Ireland_2002_Optical_Metrology_Imaging_and_Machine_Vision",

}

Murtagh, F, Qiao, X, Crookes, D, Walsh, P, Basheer, PAM & Long, A 2003, Benchmarking Segmentation Results Using a Markov Model and a Bayes Information Criterion. in A Shearer, FD Murtagh, J Mahon & PF Whelan (eds), Opto-Ireland 2002: Optical Metrology, Imaging, and Machine Vision. vol. 4877, Proceedings of SPIE - The International Society for Optical Engineering, vol. 4877, SPIE, pp. 248-254, Opto-Ireland 2002: Optical metrology, Imaging, and Machine Vision, Galway, Ireland, 5/09/02. https://doi.org/10.1117/12.467441

Benchmarking Segmentation Results Using a Markov Model and a Bayes Information Criterion. / Murtagh, Fionn; Qiao, Xiaoyu; Crookes, Danny et al.

Opto-Ireland 2002: Optical Metrology, Imaging, and Machine Vision. ed. / Andrew Shearer; Fionn D. Murtagh; James Mahon; Paul F. Whelan. Vol. 4877 SPIE, 2003. p. 248-254 (Proceedings of SPIE - The International Society for Optical Engineering; Vol. 4877).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - Benchmarking Segmentation Results Using a Markov Model and a Bayes Information Criterion

AU - Murtagh, Fionn

AU - Qiao, Xiaoyu

AU - Crookes, Danny

AU - Walsh, Paul

AU - Basheer, P. A.Muhammed

AU - Long, Adrian

PY - 2003/3/19

Y1 - 2003/3/19

N2 - Features are derived from wavelet transforms of images containing a mixture of textures. In each case, the texture mixture is segmented, based on a 10-dimensional feature vector associated with every pixel. We show that the quality of the resulting segmentations can be characterized using the Potts or Ising spatial homogeneity parameter. This measure is defined from the segmentation labels. In order to have a better measure which takes into account both the segmentation labels and the input data, we determine the likelihood of the observed data given the model, which in turn is directly related to the Bayes information criterion, BIC. Finally we discuss how BIC is used as an approximation in model assessment using a Bayes factor.

AB - Features are derived from wavelet transforms of images containing a mixture of textures. In each case, the texture mixture is segmented, based on a 10-dimensional feature vector associated with every pixel. We show that the quality of the resulting segmentations can be characterized using the Potts or Ising spatial homogeneity parameter. This measure is defined from the segmentation labels. In order to have a better measure which takes into account both the segmentation labels and the input data, we determine the likelihood of the observed data given the model, which in turn is directly related to the Bayes information criterion, BIC. Finally we discuss how BIC is used as an approximation in model assessment using a Bayes factor.

KW - Image segmentation

KW - Data modeling

KW - Expectation maximization algorithms

KW - Wavelet transforms

KW - Performance modeling

KW - Wavelets

KW - Algorithm development

KW - Visualization

KW - Civil engineering

KW - Computer science

UR - http://www.scopus.com/inward/record.url?scp=0041731699&partnerID=8YFLogxK

U2 - 10.1117/12.467441

DO - 10.1117/12.467441

M3 - Conference contribution

AN - SCOPUS:0041731699

SN - 0819446580

SN - 9780819446589

VL - 4877

T3 - Proceedings of SPIE - The International Society for Optical Engineering

SP - 248

EP - 254

BT - Opto-Ireland 2002

A2 - Shearer, Andrew

A2 - Murtagh, Fionn D.

A2 - Mahon, James

A2 - Whelan, Paul F.

PB - SPIE

T2 - Opto-Ireland 2002: Optical metrology, Imaging, and Machine Vision

Y2 - 5 September 2002 through 6 September 2002

ER -

Murtagh F, Qiao X, Crookes D, Walsh P, Basheer PAM, Long A. Benchmarking Segmentation Results Using a Markov Model and a Bayes Information Criterion. In Shearer A, Murtagh FD, Mahon J, Whelan PF, editors, Opto-Ireland 2002: Optical Metrology, Imaging, and Machine Vision. Vol. 4877. SPIE. 2003. p. 248-254. (Proceedings of SPIE - The International Society for Optical Engineering). doi: 10.1117/12.467441

Benchmarking Segmentation Results Using a Markov Model and a Bayes Information Criterion

Abstract

Publication series

Conference

Access to Document

Other files and links

Fingerprint

Cite this