| avoiding for loops [message #13445] |
Thu, 12 November 1998 00:00  |
lbryanNOSPAM
Messages: 21
Registered: July 1998
|
Junior Member |
|
|
I've had such good luck recently with y'all that I thought I'd send
another one. Although this is a little less esoteric. I need to
apply a median filter all the way to the edges of an image. So I came
up with putting an edge around it the width of the filter then
selecting out the original region. Sort of what convol does
automatically.
The following code is what I came up with. It works, but it is clunky
as heck. I know some of you guys are wizzes at getting rid of loops.
Care to lend a hand?
p.s. the data I'm working on array 4000x512 arrays not 100x100!
Thanks!!!
n = 3
bottshade = shift(dist(100),10)+randomu(100,100,100)*5.
azrange = 100
shotrange = 100
temp = fltarr(shotrange +2*n,azrange + 2*n)
temp(n,n) = bottshade
for i = 0,shotrange -1 +2*n do begin
temp(i,0:n-1) = temp(i,n)
temp(i,azrange-1+n:azrange-1+2*n) = temp(i,azrange +n - 2)
endfor
for i = 0,azrange -1 +2*n do begin
temp(0:n-1,i) = temp(n,i)
temp(shotrange-1+n:shotrange-1+2*n,i) = temp(shotrange +n - 2,i)
endfor
temp2 = median(temp,n)
bottshade = temp(n:n+shotrange-1,n:n+azrange)
end
Lisa
Arete Associates
Tucson, Arizona
lbryan@arete-az.com
|
|
|
|
| Re: Avoiding For Loops [message #45077 is a reply to message #13445] |
Mon, 01 August 2005 14:35  |
Dick Jackson
Messages: 347
Registered: August 1998
|
Senior Member |
|
|
"sudipta" <ssarkar@forestone.com> wrote in message
news:1122923758.653063.102140@f14g2000cwb.googlegroups.com.. .
> Hello Everybody,
> My IDL problem is simple but is very crucial.
. [...]
> So i was wondering if there was a way
> to avoid the for loops and do the RMS computation in one single step?
Hi Sudipta,
Your guess that IDL should be able to do this is correct. Read on! I did a
few iterations of your calculations, printing out the result, then did the
whole works after that. You can save this next section in a .pro file,
compile it and run it. Hope this helps.
=====
;; Make some sample data
data = RandomU(seed,10,14014,10,6)
lfData = RandomU(seed,6)
Print, 'Original Way:'
Print
;; I have reordered the loops with slowest-changing dimension in
;; the outer loop... in IDL, this is the last dimension.
numbands = 6
for k = 0, 1 do begin ;number of fractions (10)
for j = 0, 1 do begin ;number of spectral combinations (14013)
for i = 0,1 do begin ;number of AI (10)
tdata = reform(data(i,j,k,*),numbands,1)
Print, total((tdata - lfdata)^2)
end
Print
end
Print
end
Print, 'Un-loopy way :-) :'
Print
t0 = SysTime(1)
;; Extend the 6 values across all other dimensions
lfData = Reform(lfData,1,1,1,6)
lfDataExt = Rebin(lfData,10,14014,10,6)
;; Calculate RMS in one step
rms = Total((data-lfDataExt)^2, 4)
timeTaken = SysTime(1)-t0
Print, 'Time taken:'+StrTrim(timeTaken, 2)+' seconds.'
Print
;; Print 2x2x2 section for comparison
Print, rms[0:1,0:1,0:1]
END
=====
Printout from above program:
----------------------------
Original Way:
0.358912
1.03072
0.164605
0.782057
0.910758
0.800281
0.591216
1.48899
Un-loopy way :-) :
Time taken:0.29700017 seconds.
0.358912 1.03072
0.164605 0.782057
0.910758 0.800281
0.591216 1.48899
=====
Cheers,
--
-Dick
Dick Jackson / dick@d-jackson.com
D-Jackson Software Consulting / http://www.d-jackson.com
Calgary, Alberta, Canada / +1-403-242-7398 / Fax: 241-7392
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
