Deformable Mesh Evolved by Similarity of Image Patches

Tuan T. Nguyen; Vedrana Andersen Dahl; J. Andreas Barentzen; Anders Bjorholm Dahl

doi:10.1109/icip.2019.8803264

Deformable Mesh Evolved by Similarity of Image Patches

Tuan T. Nguyen, Vedrana Andersen Dahl, J. Andreas Barentzen, Anders Bjorholm Dahl

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

110 Downloads (Pure)

Abstract

We propose a deformable model for manually initialized segmentation of images, which may contain both textured and non-textured regions. Image segments and segment boundaries are represented using a deformable triangle mesh, providing all advantages of an explicit geometry representation, but allowing for adaptive topology. Deformation forces are computed using a probabilistic model of local self-similarity, based on clustering of image patches. Both our curve representation and our similarity model naturally support multi-label segmentation. We demonstrate the properties of our approach on a number of natural color images as well as composed textured images.

Original language	English
Title of host publication	Proceedings of 2019 IEEE International Conference on Image Processing
Publisher	IEEE
Publication date	2019
Pages	2731-2735
ISBN (Print)	978-1-5386-6249-6
DOIs	https://doi.org/10.1109/icip.2019.8803264
Publication status	Published - 2019
Event	2019 IEEE International Conference on Image Processing - Taipei International Convention Center, Taipei, Taiwan, Province of China Duration: 22 Sep 2019 → 29 Sep 2019

Conference

Conference	2019 IEEE International Conference on Image Processing
Location	Taipei International Convention Center
Country	Taiwan, Province of China
City	Taipei
Period	22/09/2019 → 29/09/2019

Keywords

Textured segmentation
Adaptive triangle mesh
Mumford-Shah

Access to Document

10.1109/icip.2019.8803264

FulltextAccepted author manuscript, 2.27 MB

Fingerprint Dive into the research topics of 'Deformable Mesh Evolved by Similarity of Image Patches'. Together they form a unique fingerprint.

Cite this

@inproceedings{a08668a89574433485f2c30414504c2f,

title = "Deformable Mesh Evolved by Similarity of Image Patches",

abstract = "We propose a deformable model for manually initialized segmentation of images, which may contain both textured and non-textured regions. Image segments and segment boundaries are represented using a deformable triangle mesh, providing all advantages of an explicit geometry representation, but allowing for adaptive topology. Deformation forces are computed using a probabilistic model of local self-similarity, based on clustering of image patches. Both our curve representation and our similarity model naturally support multi-label segmentation. We demonstrate the properties of our approach on a number of natural color images as well as composed textured images.",

keywords = "Textured segmentation, Adaptive triangle mesh, Mumford-Shah",

author = "Nguyen, {Tuan T.} and Dahl, {Vedrana Andersen} and Barentzen, {J. Andreas} and Dahl, {Anders Bjorholm}",

year = "2019",

doi = "10.1109/icip.2019.8803264",

language = "English",

isbn = "978-1-5386-6249-6",

pages = "2731--2735",

booktitle = "Proceedings of 2019 IEEE International Conference on Image Processing",

publisher = "IEEE",

address = "United States",

note = "2019 IEEE International Conference on Image Processing, ICIP 2019 ; Conference date: 22-09-2019 Through 29-09-2019",

}

TY - GEN

T1 - Deformable Mesh Evolved by Similarity of Image Patches

AU - Nguyen, Tuan T.

AU - Dahl, Vedrana Andersen

AU - Barentzen, J. Andreas

AU - Dahl, Anders Bjorholm

PY - 2019

Y1 - 2019

N2 - We propose a deformable model for manually initialized segmentation of images, which may contain both textured and non-textured regions. Image segments and segment boundaries are represented using a deformable triangle mesh, providing all advantages of an explicit geometry representation, but allowing for adaptive topology. Deformation forces are computed using a probabilistic model of local self-similarity, based on clustering of image patches. Both our curve representation and our similarity model naturally support multi-label segmentation. We demonstrate the properties of our approach on a number of natural color images as well as composed textured images.

AB - We propose a deformable model for manually initialized segmentation of images, which may contain both textured and non-textured regions. Image segments and segment boundaries are represented using a deformable triangle mesh, providing all advantages of an explicit geometry representation, but allowing for adaptive topology. Deformation forces are computed using a probabilistic model of local self-similarity, based on clustering of image patches. Both our curve representation and our similarity model naturally support multi-label segmentation. We demonstrate the properties of our approach on a number of natural color images as well as composed textured images.

KW - Textured segmentation

KW - Adaptive triangle mesh

KW - Mumford-Shah

U2 - 10.1109/icip.2019.8803264

DO - 10.1109/icip.2019.8803264

M3 - Article in proceedings

SN - 978-1-5386-6249-6

SP - 2731

EP - 2735

BT - Proceedings of 2019 IEEE International Conference on Image Processing

PB - IEEE

T2 - 2019 IEEE International Conference on Image Processing

Y2 - 22 September 2019 through 29 September 2019

ER -