| Re: iterating POLY_2D [message #19002] |
Mon, 21 February 2000 00:00 |
David Williams
Messages: 13
Registered: February 2000
|
Junior Member |
|
|
Hi again.
THe final chapter in this saga is that it was a really simple mistake!
Thanks, and sorry (respectively) to everyone who took time out to read
it and then couldn't believe I'd done something as simple as that.
Thanks especially to Christophe Morisset, who answered my posting with
remarkable speed.
Cheers.
Dave Williams
--
============================================================ =
David R. Williams, Tel.: (+44 1232) 273509
APS Division,
Pure & Applied Physics Dept.,
Queen's University,
Belfast,
BT7 1NN. http://star.pst.qub.ac.uk/~drw/
============================================================ =
|
|
|
|
| Re: iterating POLY_2D [message #19003 is a reply to message #19002] |
Mon, 21 February 2000 00:00  |
Craig Markwardt
Messages: 1869
Registered: November 1996
|
Senior Member |
|
|
David Williams <d.williams@qub.ac.uk> writes:
> ;this is the bit I'm having trouble iterating successfully:
> dataout(*,*,1)=POLY_2D(data(*,*,1),$
Shouldn't this be
> dataout(*,*,k)=POLY_2D(data(*,*,k),$
Looks like you did all the gradient calculations, but then applied
each one to the second image repeatedly.
Craig
--
------------------------------------------------------------ --------------
Craig B. Markwardt, Ph.D. EMAIL: craigmnet@cow.physics.wisc.edu
Astrophysics, IDL, Finance, Derivatives | Remove "net" for better response
------------------------------------------------------------ --------------
|
|
|
|
| Re: iterating POLY_2D [message #19012 is a reply to message #19002] |
Mon, 21 February 2000 00:00 |
David Williams
Messages: 13
Registered: February 2000
|
Junior Member |
|
|
Just as a postscript...
Christophe Morisset helpfully suggested that I also post the routine
that's giving me bother. It's included below. I'd really appreciate any
help anyone can give.
Dave Williams.
--
;+
;NAME:
; frames_move.pro
;PURPOSE:
; To remove the shaking from a sequence of digital
; images in the form of a 50x40x200 floating point
; array.
;INPUTS:
; DATA - the input array
;OUTPUTS:
; DATAOUT - the stabilised sequence
; DISPARR - the 2x200 array containing the
; displacement co-ordinates relative to
; DATA(*,*,0).
;-
PRO frames_move,data,dataout,disparr
sizD=SIZE(data)
disparr=fltarr(2,sizD(3))
dataout=fltarr(sizD(1),sizD(2),sizD(3))
dataout(*,*,0)=data(*,*,0)
;the original image will be the same in the output
;as it's the reference point from which
;the displacement in the rest of the images
;is measured.
;make the temporary gradient array to
;define the edges of the main feature:
gradex=fltarr(sizD(1),sizD(2),sizD(3))
FOR g=0,sizD(3)-1 DO BEGIN
gradex(*,*,g)=grad(data(*,*,g))
;(I've included this GRAD function
;below in a commented section. It
;comes from the IDL User's Guide.)
ENDFOR
;set up the comparison array BIGJUMP and the
;contents of the first image equal to
;the original image in the sequence passed.
bigjump=fltarr(sizD(1)-1,sizD(2)-1,2)
bigjump(*,*,0)=gradex(1:sizD(1)-1,1:sizD(2)-1,0)
FOR k=1,sizD(3)-1 DO BEGIN
bigjump(*,*,1)=gradex(1:sizD(1)-1,1:sizD(2)-1,k)
;set the second image equal to the image
;that is to be compared to the original image,
;removing the first X and Y lines which get
;wrapped around by the GRAD routine.
dxy=tr_get_disp(bigjump)
;dxy is created by a specialised routine
;written for the TRACE SolarSoft IDL distribution
disparr(0,k)=-dxy(0,1)
disparr(1,k)=-dxy(1,1)
;get the cross-correlation displacement
;this is the bit I'm having trouble iterating successfully:
dataout(*,*,1)=POLY_2D(data(*,*,1),$
[disparr(0,k),0.,1.,0.],$
[disparr(1,k),1.,0.,0.],cubic=-0.5)
PRINT,'polynomial interpolation completed for iteration
'+ARR2STR(k,/trim)
ENDFOR
END
;the GRAD.PRO function is as follows, and is used in FRAMES_MOVE.PRO
;to enhance the accuracy of the routine by detectin the edges of the
;slightly fuzzy main feature:
;
;FUNCTION GRAD, IMAGE
;
;RETURN, ABS(IMAGE - SHIFT(IMAGE,1,0)) + $
; ABS(IMAGE-SHIFT(IMAGE,0,1))
;
;END
;
============================================================ =
David R. Williams, Tel.: (+44 1232) 273509
APS Division,
Pure & Applied Physics Dept.,
Queen's University,
Belfast,
BT7 1NN. http://star.pst.qub.ac.uk/~drw/
============================================================ =
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
