| Surface spline transformation |
| Image Registration and Fusion Systems |
|
|
Surface spline, also known as thin-plate spline, is a transformation function defined in terms of logarithmic basis functions
and a linear term. The linear term makes it possible to register images with linear geometric differences exactly. Because
logarithmic functions are radially symmetric, this transformation function is particularly suitable when the landmarks are
rather uniformly spaced. If the density of landmarks varies across the image domain, large errors may be obtained away
from some landmarks. Also, since this transformation requires the solution of a system of equations, it is suitable when a
rather small set of landmarks, (fewer than 100) is available. When thousands of landmarks are available, this
transformation function not only becomes slow, the matrix of coefficients may become ill-conditioned, making it impossible
to register the images. When the landmarks are very irregularly spaced or a large number of landmarks is available,
piecewise-linear transformation is a better choice than surface spline transformation.
To learn more about this and other transformation functions for image registration, see this comprehensive report on
Transformation Functions.
Given two images and a set of corresponding landmarks in them, this software uses surface spline transformation to
resample one image to the geometry of the other.
To obtain a license for the spline transformation, follow this link =>
and a linear term. The linear term makes it possible to register images with linear geometric differences exactly. Because
logarithmic functions are radially symmetric, this transformation function is particularly suitable when the landmarks are
rather uniformly spaced. If the density of landmarks varies across the image domain, large errors may be obtained away
from some landmarks. Also, since this transformation requires the solution of a system of equations, it is suitable when a
rather small set of landmarks, (fewer than 100) is available. When thousands of landmarks are available, this
transformation function not only becomes slow, the matrix of coefficients may become ill-conditioned, making it impossible
to register the images. When the landmarks are very irregularly spaced or a large number of landmarks is available,
piecewise-linear transformation is a better choice than surface spline transformation.
To learn more about this and other transformation functions for image registration, see this comprehensive report on
Transformation Functions.
Given two images and a set of corresponding landmarks in them, this software uses surface spline transformation to
resample one image to the geometry of the other.
To obtain a license for the spline transformation, follow this link =>
This software comes in a C/C++ static library compiled with
Microsoft Visual Studio under Windows PC. The library
comes with a C/C++ driver program demonstrating usage of
the software.
Microsoft Visual Studio under Windows PC. The library
comes with a C/C++ driver program demonstrating usage of
the software.
