| Accelerating a one-line program doing matrix multiplication [message #72696] |
Mon, 27 September 2010 02:18  |
Axel Martínez
Messages: 22
Registered: June 2010
|
Junior Member |
|
|
Hi all,
I wrote a one-line function to convert a list of points from "voxel
coordinates" (image coordinates) to "real coordinates" (physical
coordinates):
;input: the points "vc", the spatial origin of an image v0 and its x,
y, and z orientation vectors (v1,v2,v3).
FUNCTION vc2rc, v0,v1,v2,v3,vc
RETURN, [[v1],[v2],[v3]] # vc + REBIN(v0, SIZE(vc, /DIMENSIONS))
END
For example, I give the image coordinate [8,1,0] and I want as output
something like [34.25, 4.12, 0], indicating the location of this voxel
in space. And the same thing but, instead of having one input point,
having several millions.
The function looks simple to me and it works great. BUT, for large
images (e.g. 500x500x200 voxels), it is terribly slow and uses way too
much memory... Am I doing something wrong, could I save speed
somewhere? I guess there should be some way to accelerate it, but I am
not able to see how...
I also have the opposite function, in my opinion also too slow (though
faster than the other)...
FUNCTION rc2vc_round, v0,v1,v2,v3,rc
RETURN, ROUND((rc - REBIN(v0, SIZE(rc, /DIMENSIONS))) ## INVERT([[v1],
[v2],[v3]]))
END
I would be really grateful for any clue!
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
