Документ взят из кэша поисковой машины. Адрес оригинального документа : http://www.stsci.edu/~mperrin/software/sources/optsubwithoffset.pro Дата изменения: Sat Dec 27 20:09:02 2008 Дата индексирования: Sat Mar 1 19:19:34 2014 Кодировка: Поисковые слова: внешние планеты

; the following is just a helper function; scroll down for the main body of optsub

FUNCTION evaloptsubwithoffset,P
common _optsub,img,psf,count,silentflag
; P is [ratio, background]

diff=img - (psf/p[0]) - P[1]

score=total(abs(diff),/nan)
count=count+1;
if (count mod 100) eq 0 then $
if not(keyword_set(silentflag)) then print,count,score,P

; message,/info,"Score is "+strc(score)
return,score


end

;+
;
; FUNCTION: optsubwithoffset
; PURPOSE:
; PSF Subtraction optimizer. Given two images already aligned,
; what scale factor optimizes their subtraction?
;
; This version also solves for a constant shift between the two images.
;
; INPUTS:
; im1,im2 2D images
;
; KEYWORDS:
; silent supress printed output
; method Registration method. See subreg.pro for documentation
; interp_type interpolation method; see StarFinder image_shift.pro
; boxsize radius of square box around star to be used
; ftol tolerance for POWELL optimization
; /show display on screen the two images
; /all use the whole image, not a subimage, for the optimization
; OUTPUTS:
; output subtracted overlap region and originals
; output[*,*,0] = subtracted ( = im1 - scale&shift(im2))
; output[*,*,1] = image 1
; output[*,*,2] = scaled and shifted image 2
;
; params a float array containing the parameters of the subtraction fit.
; scale= Divide im2 by this for best subtraction.
; offset= Add this to im2 for best subtraction.
;
; HISTORY:
; August 2001 Marshall Perrin
;
; 2006-05-25 Added /show and /all
;-

PRO optsubwithoffset,im1o,im2o,output,silent=silent,ftol=ftol,scale=scale,$
interp_type=interp_type,boxsize=boxsize,method=method,offset=offset,$
badmask = badmask,show=show,all=all
common _optsub,img,psf,count,silentflag

if not(keyword_set(ftol)) then ftol=0.0001
if not(keyword_set(boxsize)) then boxsize=30
if keyword_set(silent) then silentflag=1
sz=size(im1o)

; no registration done here! Just the scaling.

if ~(keyword_set(silent)) then message,/info,"Finding optimum subtraction, using boxsize="+strc(boxsize)
; do aperture photometry to get rough guesses at the flux ratio
im1 = im1o
im2 = im2o

; remove NANs before APER, since it chokes on them.
wnan1 = where(~ finite(im1),nanct1)
if nanct1 gt 0 then im1[wnan1] = median(im1)
wnan2 = where(~ finite(im2),nanct2)
if nanct2 gt 0 then im2[wnan2] = median(im2)

findmaxstar,im1,x1,y1,silent=silent
aper, im1, [x1], [y1], flux_img,errap,sky,skyerr,1.0,[10],[25,35],[0,0],/flux,/silent
aper, im2, [x1], [y1], flux_psf,errap,sky,skyerr,1.0,[10],[25,35],[0,0],/flux,/silent
P=[flux_psf[0]/flux_img[0],0]
if total(finite(P)) ne 2 then begin
if ~(keyword_set(silent)) then message,/info,"Aperture photometry gave NaNs; using initial guess 1.0 instead."
P = [1.0,0.0]
endif
if ~(keyword_set(silent)) then print,"Initial Fluxes are: TARGET "+strc(flux_img)+", PSF "+strc(flux_psf)+" Ratio="+strc(flux_psf[0]/flux_img[0])

; now put NANs back in to ignore those pixels.
if nanct1 gt 0 then im1[wnan1] =!values.f_nan
if nanct2 gt 0 then im2[wnan2] =!values.f_nan

if keyword_set(all) then begin
; now using entire image??!
img = im1
psf= im2
endif else begin
img=im1[0>(x1-boxsize):(x1+boxsize)< (sz[1]-1),0 >(y1-boxsize):(y1+boxsize) < (sz[2]-1)]
psf=im2[0>(x1-boxsize):(x1+boxsize)< (sz[1]-1),0 >(y1-boxsize):(y1+boxsize) < (sz[2]-1)]
endelse

if keyword_set(show) then begin
pm = !p.multi
!p.multi=[0,2,1]
imdisp,alogscale(img),title="Image 1",/axis
imdisp,alogscale(psf),title="Image 2",/axis
!p.multi=pm
endif

if keyword_set(badmask) then begin
wbad = where(badmask eq 0)
img[wbad] = !values.f_nan
psf[wbad] = !values.f_nan
endif

count=0

;if not(keyword_set(ftol)) then begin
;ftol = 1.0e-12
;endif

xi = fltarr(2,2)
xi[0,0]=.10 & xi[1,1]=.10 ;& xi[2,2]=1. & xi[3,3]=1.


; here we ignore the second parameter to P, but Powell seems to want at
; least two options available. It doesn't do anything though...

POWELL, P, xi, ftol, fmin, 'evaloptsubwithoffset',iter=iter,/double

; PRINT, 'Solution point, found after '+strc(iter)+" iterations: "
; print," : ",P[0],P[1],dxy[0],dxy[1]
; Print the value at the solution point:
if ~(keyword_set(silent)) then begin
PRINT, 'Value at solution point: ', fmin
print, "Scale factor: ",p[0]
print, "Offset : ",p[1]
endif
scale=p[0]
offset=P[1]

if arg_present(output) then begin
sub=img - (psf/p[0] ) -P[1]
sz = size(sub)
output=fltarr(sz[1],sz[2],3)

output(*,*,0)=sub
output(*,*,1)=img
output(*,*,2)=psf/p[0] +P[1]
endif

end