Hi Yakir,
On Sunday, February 09, 2014 09:14:01 PM Yakir Gagnon wrote:
> 2. regionprops
> 3. tformarray
For at least these two algorithms (and perhaps others), the only reason you
need to have them in Matlab is because looping is so dang slow (for anything
other than very specific circumstances). tformarray in particular is something
of an abomination; it always takes me 15 minutes to remind myself how to use
it. With Julia you can write much more transparent code simply by doing
something like this:
for j = 1:size(imgT, 2), i = 1:size(imgT, 1)
iT, jT = apply_my_transform(i, j)
imgT[i,j] = img[round(iT), round(jT)]
end
(for brevity this omits checks on bounds, the possibility of interpolation,
etc). This is exactly how such algorithms would be implemented if there were a
tformarray function, so this is not a "poor man's" version. On a recent Julia
build, you can even parallelize it fairly easily using SharedArrays.
> 6. interp1
This and more is in Grid.jl (it's more like griddedInterpolant, which is a
much nicer interface than interp1). I see that the docs only describe
regularly-spaced grids (which is usually the case for images), but if you look
at the code there's a bit of support for irregularly-spaced points.
Also, the restriction/prolongation code in Grid.jl is often very useful for
image processing.
--Tim
