Patch matching with polynomial exponential families and projective divergences

Frank Nielsen; Richard Nock

doi:10.1007/978-3-319-46759-7_8

Patch matching with polynomial exponential families and projective divergences

Frank Nielsen^*, Richard Nock

^*Corresponding author for this work

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

4 Citations (Scopus)

Abstract

Given a query patch image, patch matching consists in finding similar patches in a target image. In pattern recognition, patch matching is a fundamental task that is time consuming, specially when zoom factors and symmetries are handled. The matching results heavily depend on the underlying notion of distances, or similarities, between patches. We present a method that consists in modeling patches by flexible statistical parametric distributions called polynomial exponential families (PEFs). PEFs model universally arbitrary smooth distributions, and yield a compact patch representation of complexity independent of the patch sizes. Since PEFs have computationally intractable normalization terms, we estimate PEFs with score matching, and consider a projective distance: the symmetrized γ-divergence. We demonstrate experimentally the performance of our patch matching system.

Original language	English
Title of host publication	Similarity Search and Applications - 9th International Conference, SISAP 2016, Proceedings
Editors	Erich Schubert, Michael E. Houle, Laurent Amsaleg
Publisher	Springer Verlag
Pages	109-116
Number of pages	8
ISBN (Print)	9783319467580
DOIs	https://doi.org/10.1007/978-3-319-46759-7_8
Publication status	Published - 2016
Event	9th International Conference on Similarity Search and Applications, SISAP 2016 - Tokyo, Japan Duration: 24 Oct 2016 → 26 Oct 2016

Publication series

Name	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume	9939 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	9th International Conference on Similarity Search and Applications, SISAP 2016
Country/Territory	Japan
City	Tokyo
Period	24/10/16 → 26/10/16

Access to Document

10.1007/978-3-319-46759-7_8

Cite this

Nielsen, F., & Nock, R. (2016). Patch matching with polynomial exponential families and projective divergences. In E. Schubert, M. E. Houle, & L. Amsaleg (Eds.), Similarity Search and Applications - 9th International Conference, SISAP 2016, Proceedings (pp. 109-116). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 9939 LNCS). Springer Verlag. https://doi.org/10.1007/978-3-319-46759-7_8

Nielsen, Frank ; Nock, Richard. / Patch matching with polynomial exponential families and projective divergences. Similarity Search and Applications - 9th International Conference, SISAP 2016, Proceedings. editor / Erich Schubert ; Michael E. Houle ; Laurent Amsaleg. Springer Verlag, 2016. pp. 109-116 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{3251eb18fe5047c7982d54a92fec16c1,

title = "Patch matching with polynomial exponential families and projective divergences",

abstract = "Given a query patch image, patch matching consists in finding similar patches in a target image. In pattern recognition, patch matching is a fundamental task that is time consuming, specially when zoom factors and symmetries are handled. The matching results heavily depend on the underlying notion of distances, or similarities, between patches. We present a method that consists in modeling patches by flexible statistical parametric distributions called polynomial exponential families (PEFs). PEFs model universally arbitrary smooth distributions, and yield a compact patch representation of complexity independent of the patch sizes. Since PEFs have computationally intractable normalization terms, we estimate PEFs with score matching, and consider a projective distance: the symmetrized γ-divergence. We demonstrate experimentally the performance of our patch matching system.",

author = "Frank Nielsen and Richard Nock",

note = "Publisher Copyright: {\textcopyright} Springer International Publishing AG 2016.; 9th International Conference on Similarity Search and Applications, SISAP 2016 ; Conference date: 24-10-2016 Through 26-10-2016",

year = "2016",

doi = "10.1007/978-3-319-46759-7_8",

language = "English",

isbn = "9783319467580",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

publisher = "Springer Verlag",

pages = "109--116",

editor = "Erich Schubert and Houle, {Michael E.} and Laurent Amsaleg",

booktitle = "Similarity Search and Applications - 9th International Conference, SISAP 2016, Proceedings",

address = "Germany",

}

Nielsen, F & Nock, R 2016, Patch matching with polynomial exponential families and projective divergences. in E Schubert, ME Houle & L Amsaleg (eds), Similarity Search and Applications - 9th International Conference, SISAP 2016, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 9939 LNCS, Springer Verlag, pp. 109-116, 9th International Conference on Similarity Search and Applications, SISAP 2016, Tokyo, Japan, 24/10/16. https://doi.org/10.1007/978-3-319-46759-7_8

Patch matching with polynomial exponential families and projective divergences. / Nielsen, Frank; Nock, Richard.
Similarity Search and Applications - 9th International Conference, SISAP 2016, Proceedings. ed. / Erich Schubert; Michael E. Houle; Laurent Amsaleg. Springer Verlag, 2016. p. 109-116 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 9939 LNCS).

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

TY - GEN

T1 - Patch matching with polynomial exponential families and projective divergences

AU - Nielsen, Frank

AU - Nock, Richard

N1 - Publisher Copyright: © Springer International Publishing AG 2016.

PY - 2016

Y1 - 2016

N2 - Given a query patch image, patch matching consists in finding similar patches in a target image. In pattern recognition, patch matching is a fundamental task that is time consuming, specially when zoom factors and symmetries are handled. The matching results heavily depend on the underlying notion of distances, or similarities, between patches. We present a method that consists in modeling patches by flexible statistical parametric distributions called polynomial exponential families (PEFs). PEFs model universally arbitrary smooth distributions, and yield a compact patch representation of complexity independent of the patch sizes. Since PEFs have computationally intractable normalization terms, we estimate PEFs with score matching, and consider a projective distance: the symmetrized γ-divergence. We demonstrate experimentally the performance of our patch matching system.

AB - Given a query patch image, patch matching consists in finding similar patches in a target image. In pattern recognition, patch matching is a fundamental task that is time consuming, specially when zoom factors and symmetries are handled. The matching results heavily depend on the underlying notion of distances, or similarities, between patches. We present a method that consists in modeling patches by flexible statistical parametric distributions called polynomial exponential families (PEFs). PEFs model universally arbitrary smooth distributions, and yield a compact patch representation of complexity independent of the patch sizes. Since PEFs have computationally intractable normalization terms, we estimate PEFs with score matching, and consider a projective distance: the symmetrized γ-divergence. We demonstrate experimentally the performance of our patch matching system.

UR - http://www.scopus.com/inward/record.url?scp=84989963434&partnerID=8YFLogxK

U2 - 10.1007/978-3-319-46759-7_8

DO - 10.1007/978-3-319-46759-7_8

M3 - Conference contribution

SN - 9783319467580

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 109

EP - 116

BT - Similarity Search and Applications - 9th International Conference, SISAP 2016, Proceedings

A2 - Schubert, Erich

A2 - Houle, Michael E.

A2 - Amsaleg, Laurent

PB - Springer Verlag

T2 - 9th International Conference on Similarity Search and Applications, SISAP 2016

Y2 - 24 October 2016 through 26 October 2016

ER -

Nielsen F, Nock R. Patch matching with polynomial exponential families and projective divergences. In Schubert E, Houle ME, Amsaleg L, editors, Similarity Search and Applications - 9th International Conference, SISAP 2016, Proceedings. Springer Verlag. 2016. p. 109-116. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-319-46759-7_8

Patch matching with polynomial exponential families and projective divergences

Abstract

Publication series

Conference

Access to Document

Other files and links

Fingerprint

Cite this