Up: Smoothing Tasks
This task performs a 2D smoothing by an elliptical gaussian. If the in-
put file is a 3D cube (or 4D hypercube), the 2D gaussian is applied re-
peatedly on the first 2 dimensions (i.e. each velocity plane for a LMV
The algorithm works in the Fourier plane and the first 2 dimensions (in
pixels) should therefore be powers of two (but need not be equal). It is
much faster than the equivalent sky plane algorithm, but you should be-
ware of aliasing effects when the size of the convolving gaussian be-
comes a sizeable fraction of the map (say 30%).
Note also that, because of the round trip in the Fourier plane, the
blanking values can NOT be ignored. This means that each pixel with a
blanking value will contribute (pollute) the overall smoothed image in
return. In other words, you should avoid images with blank values, or at
least, set them to a reasonable dummy background value (e.g. 0).