On Apr 11, 11:05 pm, tarequea...@[EMAIL PROTECTED]
wrote:
> Hello All (IDL Gods),
>
> I am back with yet another problem.
> I know...I know...its friday night. I apologize for that. But I am
> really stuck here for a while.
>
> The problem:
>
> In a certain part of my code I need to do interpolation. The data that
> I am dealing with are from a XY grid. I need to convert them to polar
> coordinate. So, what I do is following:
>
> a. I generate a radius vector r_vec and a theta array containing
> values from zero to 2 pi.
> b. Now use the simple polar-to-rectangular coordinate transform i.e. x
> = r cos(theta) etc.
> c. Using these values I have a xy grid generated through a known r-
> theta values.
> d. Now think of superimposing the new xy grid(lets call it x'y' ) on
> to the old xy grid which contains the real data.
> e. This is the part where I need interpolation. I do interpolations to
> get the x'y' values from the xy points.
>
> So here's the question:
>
> ----- Is there any elegant way of doing this coordinate
> transformation ? (And in case you are thinking, "well you already have
> the xy data, so why not just convert to r-theta?", I have to say that
> the interpolation method actually gives me a way nicer dataset ).
>
> My 2nd trouble is, and this is probably the biggest and dumbest
> problem for me.
>
> ----- i was playing around with several interpolation routines from
> IDl. My boss's suggestion was to use 'bilinear'.
> but I thought to give others' a shot too. Problem is, when I am
> done with interpolation, result is nothing like what I was
> expecting. A run down version of the code is shown below:
>
> =======================start of code================================
>
> Nth= 10.
> dth= 1/Nth
> r_vec= findgen(Nth)/Nth
> theta_vec = findgen(Nth)/Nth * 2.*!Pi
>
> for i=0L,Nth - 1 do begin
>
> x[i]= r_vec[i]* cos(theta_vec)
>
> endfor
>
> for j=0L,Nth - 1 do begin
>
> y[j]= r_vec[j]* sin(theta_vec)
>
> endfor
>
> ;print,y
>
> ;plot,x,y
>
> -----------------------------------------------------------------------
> Now I create the 'main' dataset on which I am going to use
> interpolation scheme.
>
> rr = findgen(20.)/30.
> tht = findgen(20.)/30. *2*!Pi
>
> m = fltarr(20,20)
>
> for j=0,19 do begin
> for i=0,19 do begin
>
> m[i,j] = rr[i]*cos(tht[j]) + 5.*rr[i]*sin(tht[j])
>
> ;print,i
> endfor
> endfor
>
> m_p=bilinear(m,x,y)
>
> End
> =======================End of Code=================================
>
> The problem is, as I mentioned above, when I plot m and the
> interpolated m_p, they do not look like similar at all.
>
> Any help will be greatly appreciated.
>
> Thanks in advance.
>
> ~tareque
Hi,
Coordinate transforms are very easy with CV_COORD().
I don't understand where you want to go with the interpolation. The
input coordinates to BILINEAR are rectangular and formed as indexed
based (as in subscript indices in X and Y.) As near as I can
reconstruct, X ranges from -0.500000 to 0.728115 and Y ranges
from -0.760845 to 0.285317. My reconstruction of your code is
below.
Cheers,
Ben
PRO tareque
Nth= 10L
dth= 1.0/Nth
r_vec= findgen(Nth)/Nth
theta_vec = findgen(Nth)/Nth * 2.*!Pi
x = fltarr(nth)
y = fltarr(nth)
for i=0L,Nth - 1 do begin
x[i]= r_vec[i]* cos(theta_vec[i])
endfor
for j=0L,Nth - 1 do begin
y[j]= r_vec[j]* sin(theta_vec[j])
endfor
;print,y
;plot,x,y
;-----------------------------------------------------------------------
;Now I create the 'main' dataset on which I am going to use
;interpolation scheme.
rr = findgen(20.)/30.
tht = findgen(20.)/30. *2*!Pi
m = fltarr(20,20)
for j=0,19 do begin
for i=0,19 do begin
m[i,j] = rr[i]*cos(tht[j]) + 5.*rr[i]*sin(tht[j])
;print,i
endfor
endfor
m_p=bilinear(m,x,y)
STOP
End
|
