| Re: Finger Pointing in IDL [message #53408] |
Sat, 14 April 2007 09:02  |
David Fanning
Messages: 11724
Registered: August 2001
|
Senior Member |
|
|
David Fanning writes:
> Should I fix it in my programs (lots of work), ask Craig
> to fix it in his (less work, at least for me), or request
> ITTVIS make CONGRID work in a manner that we would all
> prefer it would work (probably less work, since they could
> copy Craig's fixes for their program, and I could explain to
> them how to fix my current problem)?
>
> I thought it would be interesting to hear your opinion.
OK, I understand, it's MY problem, not YOURS! But, still,
someone must have an idea about this. In any case, *I've*
been giving it more thought.
I've been thinking that a non-interpolated solution in
3D maybe doesn't make any sense in general. Maybe that's
why neither CONGRID or CMCONGRID allow it. Certainly
you could do 2D non-interpolated solutions over the
third dimension, as someone suggested the other day,
but it seems to be that which dimension you select
would be arbitrary, and it is not clear to me you would
end up with the same solution in different directions.
Of course, a 24-bit image is a fairly specific example
of the general 3D case, so looping over the interleaved
dimension could make some sense in this case. Which
would make it MY problem again. Humm.
Well, just thinking out loud in the absence of other
conversation... :-)
Cheers,
David
--
David Fanning, Ph.D.
Fanning Software Consulting, Inc.
Coyote's Guide to IDL Programming: http://www.dfanning.com/
Sepore ma de ni thui. ("Perhaps thou speakest truth.")
|
|
|
|
| Re: Finger Pointing in IDL [message #53530 is a reply to message #53408] |
Mon, 16 April 2007 08:30  |
edward.s.meinel@aero.
Messages: 52
Registered: February 2005
|
Member |
|
|
On Apr 14, 12:02 pm, David Fanning <n...@dfanning.com> wrote:
>
> Of course, a 24-bit image is a fairly specific example
> of the general 3D case, so looping over the interleaved
> dimension could make some sense in this case. Which
> would make it MY problem again. Humm.
What? You can't use NN on color images? That's just dumb...
> it is not clear to me you would
> end up with the same solution in different directions.
It isn't clear to me why you wouldn't end up with the same solution in
different directions.
Ed
|
|
|
|
| Re: Finger Pointing in IDL [message #53536 is a reply to message #53408] |
Mon, 16 April 2007 04:48 |
Qing
Messages: 12
Registered: February 2007
|
Junior Member |
|
|
On Apr 15, 2:02 am, David Fanning <n...@dfanning.com> wrote:
> David Fanning writes:
>> Should I fix it in my programs (lots of work), ask Craig
>> to fix it in his (less work, at least for me), or request
>> ITTVIS make CONGRID work in a manner that we would all
>> prefer it would work (probably less work, since they could
>> copy Craig's fixes for their program, and I could explain to
>> them how to fix my current problem)?
>
>> I thought it would be interesting to hear your opinion.
>
> OK, I understand, it's MY problem, not YOURS! But, still,
> someone must have an idea about this. In any case, *I've*
> been giving it more thought.
>
> I've been thinking that a non-interpolated solution in
> 3D maybe doesn't make any sense in general. Maybe that's
> why neither CONGRID or CMCONGRID allow it. Certainly
> you could do 2D non-interpolated solutions over the
> third dimension, as someone suggested the other day,
> but it seems to be that which dimension you select
> would be arbitrary, and it is not clear to me you would
> end up with the same solution in different directions.
Hi David.
A very important point! But would be interesting if someone
can find a case to give the different results!
Cheers.
Qing
|
|
|
|
| Re: Finger Pointing in IDL [message #53537 is a reply to message #53408] |
Mon, 16 April 2007 04:42 |
Qing
Messages: 12
Registered: February 2007
|
Junior Member |
|
|
On Apr 15, 2:02 am, David Fanning <n...@dfanning.com> wrote:
> David Fanning writes:
>> Should I fix it in my programs (lots of work), ask Craig
>> to fix it in his (less work, at least for me), or request
>> ITTVIS make CONGRID work in a manner that we would all
>> prefer it would work (probably less work, since they could
>> copy Craig's fixes for their program, and I could explain to
>> them how to fix my current problem)?
>
>> I thought it would be interesting to hear your opinion.
>
> OK, I understand, it's MY problem, not YOURS! But, still,
> someone must have an idea about this. In any case, *I've*
> been giving it more thought.
>
> I've been thinking that a non-interpolated solution in
> 3D maybe doesn't make any sense in general. Maybe that's
> why neither CONGRID or CMCONGRID allow it. Certainly
> you could do 2D non-interpolated solutions over the
> third dimension, as someone suggested the other day,
> but it seems to be that which dimension you select
> would be arbitrary, and it is not clear to me you would
> end up with the same solution in different directions.
>
Hi David,
That's an important point - does it matter to select which
dimension to loop over? Based on Michael Galloy's analysis
in the original lead on this issue, it seems to me that there
is only one target solution (without interpolation).
A fair test would be to find a case that gives different
results depending on the selection of which dimention.
It will be interesting if someone can find such a case!
Cheers.
Qing
|
|
|
|
| Re: Finger Pointing in IDL [message #53538 is a reply to message #53408] |
Mon, 16 April 2007 04:00 |
Jo Klein
Messages: 54
Registered: January 2006
|
Member |
|
|
Hi David,
I wouldn't say 3D nearest neighbour generally doesn't make sense - it's
useful when you resize masks that should be used to derive some
measurement or other from a different volume. Typical example: Mask
drawn on volume MRI, which is then transported to some other modality,
e.g. fMRI or PET. Many imaging packages won't take fractional values for
such masks, so nearest neighbour is the way to go.
Cheers,
Jo
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
