Improved optimization methods for image registration problems

Ke Chen; Geovani Nunes Grapiglia; Jinyun Yuan; Daoping Zhang

doi:10.1007/s11075-018-0486-2

Improved optimization methods for image registration problems

Ke Chen^*, Geovani Nunes Grapiglia, Jinyun Yuan, Daoping Zhang

^*Corresponding author for this work

Mathematics And Statistics

Research output: Contribution to journal › Article › peer-review

7 Citations (Scopus)

9 Downloads (Pure)

Abstract

In this paper, we propose new multilevel optimization methods for minimizing continuously differentiable functions obtained by discretizing models for image registration problems. These multilevel schemes rely on a novel two-step Gauss-Newton method, in which a second step is computed within each iteration by minimizing a quadratic approximation of the objective function over a certain two-dimensional subspace. Numerical results on image registration problems show that the proposed methods can outperform the standard multilevel Gauss-Newton method.

Original language	English
Pages (from-to)	305-336
Number of pages	32
Journal	Numerical Algorithms
Volume	80
Issue number	2
DOIs	https://doi.org/10.1007/s11075-018-0486-2
Publication status	Published - 6 Feb 2019

Funding

Funding information This work was partially supported by UK EPSRC (grants EP/K036939/1 and EP/N014499/1), by the Newton Research Collaboration Programme (grant NRCP 1617/6/187) and by the National Council for Scientific and Technological Development (grant CNPq 406269/2016-5).

Keywords

Gauss-Newton method
image registration
multilevel strategy
subspace methods

Access to Document

10.1007/s11075-018-0486-2

Chen-etal-NA-2019-Improved-optimization-methods-for-image-registration-problemsAccepted author manuscript, 10.2 MBLicence: Strath_1

Cite this

@article{61b12d768ae4417d8604d9be74c08c0c,

title = "Improved optimization methods for image registration problems",

abstract = "In this paper, we propose new multilevel optimization methods for minimizing continuously differentiable functions obtained by discretizing models for image registration problems. These multilevel schemes rely on a novel two-step Gauss-Newton method, in which a second step is computed within each iteration by minimizing a quadratic approximation of the objective function over a certain two-dimensional subspace. Numerical results on image registration problems show that the proposed methods can outperform the standard multilevel Gauss-Newton method.",

keywords = "Gauss-Newton method, image registration, multilevel strategy, subspace methods",

author = "Ke Chen and Grapiglia, {Geovani Nunes} and Jinyun Yuan and Daoping Zhang",

year = "2019",

month = feb,

day = "6",

doi = "10.1007/s11075-018-0486-2",

language = "English",

volume = "80",

pages = "305--336",

journal = "Numerical Algorithms",

issn = "1017-1398",

publisher = "Springer Netherlands",

number = "2",

}

TY - JOUR

T1 - Improved optimization methods for image registration problems

AU - Chen, Ke

AU - Grapiglia, Geovani Nunes

AU - Yuan, Jinyun

AU - Zhang, Daoping

PY - 2019/2/6

Y1 - 2019/2/6

N2 - In this paper, we propose new multilevel optimization methods for minimizing continuously differentiable functions obtained by discretizing models for image registration problems. These multilevel schemes rely on a novel two-step Gauss-Newton method, in which a second step is computed within each iteration by minimizing a quadratic approximation of the objective function over a certain two-dimensional subspace. Numerical results on image registration problems show that the proposed methods can outperform the standard multilevel Gauss-Newton method.

AB - In this paper, we propose new multilevel optimization methods for minimizing continuously differentiable functions obtained by discretizing models for image registration problems. These multilevel schemes rely on a novel two-step Gauss-Newton method, in which a second step is computed within each iteration by minimizing a quadratic approximation of the objective function over a certain two-dimensional subspace. Numerical results on image registration problems show that the proposed methods can outperform the standard multilevel Gauss-Newton method.

KW - Gauss-Newton method

KW - image registration

KW - multilevel strategy

KW - subspace methods

UR - https://livrepository.liverpool.ac.uk/3018792/

U2 - 10.1007/s11075-018-0486-2

DO - 10.1007/s11075-018-0486-2

M3 - Article

AN - SCOPUS:85042443072

SN - 1017-1398

VL - 80

SP - 305

EP - 336

JO - Numerical Algorithms

JF - Numerical Algorithms

IS - 2

ER -

Improved optimization methods for image registration problems

Abstract

Funding

Keywords

Access to Document

Other files and links

Fingerprint

Cite this