Corrected laplacians

Closer cuts and segmentation with shape priors

David Tolliver, Gary L. Miller, Robert Collins

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

4 Citations (Scopus)

Abstract

We optimize over the set of corrected laplacians (CL) associated with a weighted graph to improve the average case normalized cut (NCut) of a graph. Unlike edge-relaxation SDPs, optimizing over the set CL naturally exploits the matrix sparsity by operating solely on the diagonal. This structure is critical to image segmentation applications because the number of vertices is generally proportional to the number of pixels in the image. CL optimization provides a guiding principle for improving the combinatorial solution over the spectral relaxation, which is important because small improvements in the cut cost often result in significant improvements in the perceptual relevance of the segmentation. We develop an optimization procedure to accommodate prior information in the form of statistical shape models, resulting in a segmentation method that produces foreground regions which are consistent with a parameterized family of shapes. We validate our technique with ground truth on MRI medical images, providing a quantitative comparison against results produced by current spectral relaxation approaches to graph partitioning.

Original languageEnglish (US)
Title of host publicationProceedings - 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, CVPR 2005
PublisherIEEE Computer Society
Pages92-98
Number of pages7
ISBN (Print)0769523722, 9780769523729
DOIs
StatePublished - Jan 1 2005
Event2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, CVPR 2005 - San Diego, CA, United States
Duration: Jun 20 2005Jun 25 2005

Publication series

NameProceedings - 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, CVPR 2005
VolumeII

Other

Other2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, CVPR 2005
CountryUnited States
CitySan Diego, CA
Period6/20/056/25/05

Fingerprint

Image segmentation
Magnetic resonance imaging
Pixels
Costs

All Science Journal Classification (ASJC) codes

  • Engineering(all)

Cite this

Tolliver, D., Miller, G. L., & Collins, R. (2005). Corrected laplacians: Closer cuts and segmentation with shape priors. In Proceedings - 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, CVPR 2005 (pp. 92-98). [1467427] (Proceedings - 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, CVPR 2005; Vol. II). IEEE Computer Society. https://doi.org/10.1109/CVPR.2005.112
Tolliver, David ; Miller, Gary L. ; Collins, Robert. / Corrected laplacians : Closer cuts and segmentation with shape priors. Proceedings - 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, CVPR 2005. IEEE Computer Society, 2005. pp. 92-98 (Proceedings - 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, CVPR 2005).
@inproceedings{e08f395f92d04dde93d4c3f34073825e,
title = "Corrected laplacians: Closer cuts and segmentation with shape priors",
abstract = "We optimize over the set of corrected laplacians (CL) associated with a weighted graph to improve the average case normalized cut (NCut) of a graph. Unlike edge-relaxation SDPs, optimizing over the set CL naturally exploits the matrix sparsity by operating solely on the diagonal. This structure is critical to image segmentation applications because the number of vertices is generally proportional to the number of pixels in the image. CL optimization provides a guiding principle for improving the combinatorial solution over the spectral relaxation, which is important because small improvements in the cut cost often result in significant improvements in the perceptual relevance of the segmentation. We develop an optimization procedure to accommodate prior information in the form of statistical shape models, resulting in a segmentation method that produces foreground regions which are consistent with a parameterized family of shapes. We validate our technique with ground truth on MRI medical images, providing a quantitative comparison against results produced by current spectral relaxation approaches to graph partitioning.",
author = "David Tolliver and Miller, {Gary L.} and Robert Collins",
year = "2005",
month = "1",
day = "1",
doi = "10.1109/CVPR.2005.112",
language = "English (US)",
isbn = "0769523722",
series = "Proceedings - 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, CVPR 2005",
publisher = "IEEE Computer Society",
pages = "92--98",
booktitle = "Proceedings - 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, CVPR 2005",
address = "United States",

}

Tolliver, D, Miller, GL & Collins, R 2005, Corrected laplacians: Closer cuts and segmentation with shape priors. in Proceedings - 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, CVPR 2005., 1467427, Proceedings - 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, CVPR 2005, vol. II, IEEE Computer Society, pp. 92-98, 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, CVPR 2005, San Diego, CA, United States, 6/20/05. https://doi.org/10.1109/CVPR.2005.112

Corrected laplacians : Closer cuts and segmentation with shape priors. / Tolliver, David; Miller, Gary L.; Collins, Robert.

Proceedings - 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, CVPR 2005. IEEE Computer Society, 2005. p. 92-98 1467427 (Proceedings - 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, CVPR 2005; Vol. II).

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

TY - GEN

T1 - Corrected laplacians

T2 - Closer cuts and segmentation with shape priors

AU - Tolliver, David

AU - Miller, Gary L.

AU - Collins, Robert

PY - 2005/1/1

Y1 - 2005/1/1

N2 - We optimize over the set of corrected laplacians (CL) associated with a weighted graph to improve the average case normalized cut (NCut) of a graph. Unlike edge-relaxation SDPs, optimizing over the set CL naturally exploits the matrix sparsity by operating solely on the diagonal. This structure is critical to image segmentation applications because the number of vertices is generally proportional to the number of pixels in the image. CL optimization provides a guiding principle for improving the combinatorial solution over the spectral relaxation, which is important because small improvements in the cut cost often result in significant improvements in the perceptual relevance of the segmentation. We develop an optimization procedure to accommodate prior information in the form of statistical shape models, resulting in a segmentation method that produces foreground regions which are consistent with a parameterized family of shapes. We validate our technique with ground truth on MRI medical images, providing a quantitative comparison against results produced by current spectral relaxation approaches to graph partitioning.

AB - We optimize over the set of corrected laplacians (CL) associated with a weighted graph to improve the average case normalized cut (NCut) of a graph. Unlike edge-relaxation SDPs, optimizing over the set CL naturally exploits the matrix sparsity by operating solely on the diagonal. This structure is critical to image segmentation applications because the number of vertices is generally proportional to the number of pixels in the image. CL optimization provides a guiding principle for improving the combinatorial solution over the spectral relaxation, which is important because small improvements in the cut cost often result in significant improvements in the perceptual relevance of the segmentation. We develop an optimization procedure to accommodate prior information in the form of statistical shape models, resulting in a segmentation method that produces foreground regions which are consistent with a parameterized family of shapes. We validate our technique with ground truth on MRI medical images, providing a quantitative comparison against results produced by current spectral relaxation approaches to graph partitioning.

UR - http://www.scopus.com/inward/record.url?scp=24644450300&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=24644450300&partnerID=8YFLogxK

U2 - 10.1109/CVPR.2005.112

DO - 10.1109/CVPR.2005.112

M3 - Conference contribution

SN - 0769523722

SN - 9780769523729

T3 - Proceedings - 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, CVPR 2005

SP - 92

EP - 98

BT - Proceedings - 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, CVPR 2005

PB - IEEE Computer Society

ER -

Tolliver D, Miller GL, Collins R. Corrected laplacians: Closer cuts and segmentation with shape priors. In Proceedings - 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, CVPR 2005. IEEE Computer Society. 2005. p. 92-98. 1467427. (Proceedings - 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition, CVPR 2005). https://doi.org/10.1109/CVPR.2005.112