[Insight-users] spline as a combination of b-splines

[Amy C]

Hello,

I'm after some sort of B-spline class, but I'm not sure what to use.
I have a spline that is a linear combination of cubic (uniform, centered)
b-splines, it maps from R^(image dimension) -> R.  ( spline[px] = sum(
coefficient[px]*b-spline[px] ).
In this case, instead of interpolating a set of numbers, I -know- the
coefficients(/control points) and knots of the spline. More or less, each
pixel in the image contains a coefficient/control point (or every n pixels,
where n is an integer) and I'd like to evaluate the spline.
I guess I just need some object with the following properties
- set coefficients(/control points)
- set knots ( or spacing? )
- evaluate the spline at every pixel
- calculate the derivative
- will be used in a loop - will recalculate coefficients each loop
- need to convolve the coefficient 'image' and possibly downsample every now
and then

At first I thought of just using a normal array to hold the coefficients
etc, but the last two points suggest I should use something within ITK. I
was thinking of making an object that is more or less an itk::Image (for the
coefficients) with an Evaluate/EvaluateDerivative method.

Does anyone know of a class that does this already? (Set knots, set
coefficients, evaluate both spline and derivatives, where the spline is made
up of bsplines?) BSplineInterpolateImageFunction seems closest, but it
calculates coefficients to fit the input image, while I know the
coefficients and just want to calculate the derivative/values.

Thanks,
Amy
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.itk.org/pipermail/insight-users/attachments/20100103/4a2bc416/attachment.htm>