Fast interpolation operations in non-rigid image registration

Matthew W. Jacobson, Jeffrey A. Fessler

Research output: Contribution to journalArticle

Abstract

Much literature on image registration-13 has worked with purely geometric image deformation models. For such models, interpolation/resampling operations are often the computationally intensive steps when iteratively minimizing the deformation cost function. This article discusses some techniques for efficiently implementing and accelerating these operations. To simplify presentation, we discuss our ideas in the context of 2D imaging. However, the concepts readily generalize to 3D. Our central technique is a table-lookup scheme that makes somewhat liberal use of RAM, but should not strain the resources of modern processors if certain design parameters are appropriately selected. The technique works by preinterpolating and tabulating the grid values of the reference image onto a finer grid along one of the axes of the image. The lookup table can be rapidly constructed using FFTs. Our results show that this technique reduces iterative computation by an order of magnitude. When a minimization algorithm employing coordinate block alternation is used, one can obtain still faster computation by storing certain intermediate quantities as state variables. We refer to this technique as state variable hold-over. When combined with table-lookup, state variable hold-over reduces CPU time by about a factor two, as compared to table-lookup alone.

Original languageEnglish (US)
Article number19
Pages (from-to)764-774
Number of pages11
JournalProgress in Biomedical Optics and Imaging - Proceedings of SPIE
Volume5747
Issue numberII
DOIs
StatePublished - 2005
Externally publishedYes

Fingerprint

Table lookup
Image registration
interpolation
Interpolation
Reference Values
Costs and Cost Analysis
tabulation processes
grids
fast Fourier transformations
alternations
Random access storage
Cost functions
Fast Fourier transforms
Program processors
central processing units
resources
costs
Imaging techniques
optimization

ASJC Scopus subject areas

  • Engineering(all)

Cite this

Fast interpolation operations in non-rigid image registration. / Jacobson, Matthew W.; Fessler, Jeffrey A.

In: Progress in Biomedical Optics and Imaging - Proceedings of SPIE, Vol. 5747, No. II, 19, 2005, p. 764-774.

Research output: Contribution to journalArticle

Jacobson, Matthew W. ; Fessler, Jeffrey A. / Fast interpolation operations in non-rigid image registration. In: Progress in Biomedical Optics and Imaging - Proceedings of SPIE. 2005 ; Vol. 5747, No. II. pp. 764-774.
@article{a701d73482414b0c898fd397ad69cb5c,
title = "Fast interpolation operations in non-rigid image registration",
abstract = "Much literature on image registration-13 has worked with purely geometric image deformation models. For such models, interpolation/resampling operations are often the computationally intensive steps when iteratively minimizing the deformation cost function. This article discusses some techniques for efficiently implementing and accelerating these operations. To simplify presentation, we discuss our ideas in the context of 2D imaging. However, the concepts readily generalize to 3D. Our central technique is a table-lookup scheme that makes somewhat liberal use of RAM, but should not strain the resources of modern processors if certain design parameters are appropriately selected. The technique works by preinterpolating and tabulating the grid values of the reference image onto a finer grid along one of the axes of the image. The lookup table can be rapidly constructed using FFTs. Our results show that this technique reduces iterative computation by an order of magnitude. When a minimization algorithm employing coordinate block alternation is used, one can obtain still faster computation by storing certain intermediate quantities as state variables. We refer to this technique as state variable hold-over. When combined with table-lookup, state variable hold-over reduces CPU time by about a factor two, as compared to table-lookup alone.",
author = "Jacobson, {Matthew W.} and Fessler, {Jeffrey A.}",
year = "2005",
doi = "10.1117/12.592243",
language = "English (US)",
volume = "5747",
pages = "764--774",
journal = "Progress in Biomedical Optics and Imaging - Proceedings of SPIE",
issn = "1605-7422",
publisher = "SPIE",
number = "II",

}

TY - JOUR

T1 - Fast interpolation operations in non-rigid image registration

AU - Jacobson, Matthew W.

AU - Fessler, Jeffrey A.

PY - 2005

Y1 - 2005

N2 - Much literature on image registration-13 has worked with purely geometric image deformation models. For such models, interpolation/resampling operations are often the computationally intensive steps when iteratively minimizing the deformation cost function. This article discusses some techniques for efficiently implementing and accelerating these operations. To simplify presentation, we discuss our ideas in the context of 2D imaging. However, the concepts readily generalize to 3D. Our central technique is a table-lookup scheme that makes somewhat liberal use of RAM, but should not strain the resources of modern processors if certain design parameters are appropriately selected. The technique works by preinterpolating and tabulating the grid values of the reference image onto a finer grid along one of the axes of the image. The lookup table can be rapidly constructed using FFTs. Our results show that this technique reduces iterative computation by an order of magnitude. When a minimization algorithm employing coordinate block alternation is used, one can obtain still faster computation by storing certain intermediate quantities as state variables. We refer to this technique as state variable hold-over. When combined with table-lookup, state variable hold-over reduces CPU time by about a factor two, as compared to table-lookup alone.

AB - Much literature on image registration-13 has worked with purely geometric image deformation models. For such models, interpolation/resampling operations are often the computationally intensive steps when iteratively minimizing the deformation cost function. This article discusses some techniques for efficiently implementing and accelerating these operations. To simplify presentation, we discuss our ideas in the context of 2D imaging. However, the concepts readily generalize to 3D. Our central technique is a table-lookup scheme that makes somewhat liberal use of RAM, but should not strain the resources of modern processors if certain design parameters are appropriately selected. The technique works by preinterpolating and tabulating the grid values of the reference image onto a finer grid along one of the axes of the image. The lookup table can be rapidly constructed using FFTs. Our results show that this technique reduces iterative computation by an order of magnitude. When a minimization algorithm employing coordinate block alternation is used, one can obtain still faster computation by storing certain intermediate quantities as state variables. We refer to this technique as state variable hold-over. When combined with table-lookup, state variable hold-over reduces CPU time by about a factor two, as compared to table-lookup alone.

UR - http://www.scopus.com/inward/record.url?scp=23844497144&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=23844497144&partnerID=8YFLogxK

U2 - 10.1117/12.592243

DO - 10.1117/12.592243

M3 - Article

AN - SCOPUS:23844497144

VL - 5747

SP - 764

EP - 774

JO - Progress in Biomedical Optics and Imaging - Proceedings of SPIE

JF - Progress in Biomedical Optics and Imaging - Proceedings of SPIE

SN - 1605-7422

IS - II

M1 - 19

ER -