Generic maximum likely scale selection

Kim Steenstrup Pedersen; Marco Loog; Bo Markussen

doi:10.1007/978-3-540-72823-8_31

Generic maximum likely scale selection

Kim Steenstrup Pedersen, Marco Loog, Bo Markussen

Department of Computer Science

1 Citation (Scopus)

Abstract

The fundamental problem of local scale selection is addressed by
means of a novel principle, which is based on maximum likelihood
estimation. The principle is generally applicable to a broad
variety of image models and descriptors, and provides a generic
scale estimation methodology.

The focus in this work is on applying this selection principle
under a Brownian image model. This image model provides a simple
scale invariant prior for natural images and we provide
illustrative examples of the behavior of our scale estimation on
such images. In these illustrative examples, estimation is based
on second order moments of multiple measurements outputs at a
fixed location. These measurements, which reflect local image
structure, consist in the cases considered here of Gaussian
derivatives taken at several scales and/or having different
derivative orders.

Original language	English
Title of host publication	Scale Space and Variational Methods in Computer Vision : First International conference , SSVM 2007, Ischia, Italy, May 30 - June 2, 2007. Proceedings
Editors	Fiorella Sgallari, Almerica Murli, Nikos Paragios
Number of pages	12
Publisher	Springer
Publication date	2007
Pages	362-373
ISBN (Print)	978-3-540-72822-1
DOIs	https://doi.org/10.1007/978-3-540-72823-8_31
Publication status	Published - 2007
Event	International Conference on Scale Space and Variational Methods in Computer Vision (SSVM 2007) - Ischia, Italy Duration: 30 May 2007 → 2 Jun 2007 Conference number: 1

Conference

Conference	International Conference on Scale Space and Variational Methods in Computer Vision (SSVM 2007)
Number	1
Country/Territory	Italy
City	Ischia
Period	30/05/2007 → 02/06/2007

Series	Lecture notes in computer science
Number	4485
ISSN	0302-9743

Access to Document

10.1007/978-3-540-72823-8_31

Cite this

Generic maximum likely scale selection. / Pedersen, Kim Steenstrup; Loog, Marco; Markussen, Bo.
Scale Space and Variational Methods in Computer Vision: First International conference , SSVM 2007, Ischia, Italy, May 30 - June 2, 2007. Proceedings. ed. / Fiorella Sgallari; Almerica Murli; Nikos Paragios. Springer, 2007. p. 362-373 (Lecture notes in computer science; No. 4485).

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

Pedersen, KS, Loog, M & Markussen, B 2007, Generic maximum likely scale selection. in F Sgallari, A Murli & N Paragios (eds), Scale Space and Variational Methods in Computer Vision: First International conference , SSVM 2007, Ischia, Italy, May 30 - June 2, 2007. Proceedings. Springer, Lecture notes in computer science, no. 4485, pp. 362-373, International Conference on Scale Space and Variational Methods in Computer Vision (SSVM 2007), Ischia, Italy, 30/05/2007. https://doi.org/10.1007/978-3-540-72823-8_31

Pedersen KS, Loog M, Markussen B. Generic maximum likely scale selection. In Sgallari F, Murli A, Paragios N, editors, Scale Space and Variational Methods in Computer Vision: First International conference , SSVM 2007, Ischia, Italy, May 30 - June 2, 2007. Proceedings. Springer. 2007. p. 362-373. (Lecture notes in computer science; No. 4485). doi: 10.1007/978-3-540-72823-8_31

Pedersen, Kim Steenstrup ; Loog, Marco ; Markussen, Bo. / Generic maximum likely scale selection. Scale Space and Variational Methods in Computer Vision: First International conference , SSVM 2007, Ischia, Italy, May 30 - June 2, 2007. Proceedings. editor / Fiorella Sgallari ; Almerica Murli ; Nikos Paragios. Springer, 2007. pp. 362-373 (Lecture notes in computer science; No. 4485).

@inproceedings{779074d0b54c11dcbee902004c4f4f50,

title = "Generic maximum likely scale selection",

abstract = "The fundamental problem of local scale selection is addressed bymeans of a novel principle, which is based on maximum likelihoodestimation. The principle is generally applicable to a broadvariety of image models and descriptors, and provides a genericscale estimation methodology.The focus in this work is on applying this selection principleunder a Brownian image model. This image model provides a simplescale invariant prior for natural images and we provideillustrative examples of the behavior of our scale estimation onsuch images. In these illustrative examples, estimation is basedon second order moments of multiple measurements outputs at afixed location. These measurements, which reflect local imagestructure, consist in the cases considered here of Gaussianderivatives taken at several scales and/or having differentderivative orders.",

author = "Pedersen, {Kim Steenstrup} and Marco Loog and Bo Markussen",

year = "2007",

doi = "10.1007/978-3-540-72823-8_31",

language = "English",

isbn = "978-3-540-72822-1",

series = "Lecture notes in computer science",

publisher = "Springer",

number = "4485",

pages = "362--373",

editor = "Fiorella Sgallari and Almerica Murli and Nikos Paragios",

booktitle = "Scale Space and Variational Methods in Computer Vision",

note = " International Conference on Scale Space and Variational Methods in Computer Vision (SSVM 2007) ; Conference date: 30-05-2007 Through 02-06-2007",

}

TY - GEN

T1 - Generic maximum likely scale selection

AU - Pedersen, Kim Steenstrup

AU - Loog, Marco

AU - Markussen, Bo

N1 - Conference code: 1

PY - 2007

Y1 - 2007

N2 - The fundamental problem of local scale selection is addressed bymeans of a novel principle, which is based on maximum likelihoodestimation. The principle is generally applicable to a broadvariety of image models and descriptors, and provides a genericscale estimation methodology.The focus in this work is on applying this selection principleunder a Brownian image model. This image model provides a simplescale invariant prior for natural images and we provideillustrative examples of the behavior of our scale estimation onsuch images. In these illustrative examples, estimation is basedon second order moments of multiple measurements outputs at afixed location. These measurements, which reflect local imagestructure, consist in the cases considered here of Gaussianderivatives taken at several scales and/or having differentderivative orders.

AB - The fundamental problem of local scale selection is addressed bymeans of a novel principle, which is based on maximum likelihoodestimation. The principle is generally applicable to a broadvariety of image models and descriptors, and provides a genericscale estimation methodology.The focus in this work is on applying this selection principleunder a Brownian image model. This image model provides a simplescale invariant prior for natural images and we provideillustrative examples of the behavior of our scale estimation onsuch images. In these illustrative examples, estimation is basedon second order moments of multiple measurements outputs at afixed location. These measurements, which reflect local imagestructure, consist in the cases considered here of Gaussianderivatives taken at several scales and/or having differentderivative orders.

U2 - 10.1007/978-3-540-72823-8_31

DO - 10.1007/978-3-540-72823-8_31

M3 - Article in proceedings

SN - 978-3-540-72822-1

T3 - Lecture notes in computer science

SP - 362

EP - 373

BT - Scale Space and Variational Methods in Computer Vision

A2 - Sgallari, Fiorella

A2 - Murli, Almerica

A2 - Paragios, Nikos

PB - Springer

T2 - International Conference on Scale Space and Variational Methods in Computer Vision (SSVM 2007)

Y2 - 30 May 2007 through 2 June 2007

ER -

Generic maximum likely scale selection

Abstract

Conference

Access to Document

Fingerprint

Cite this