function sxt_vignette, index, energy, qtest=qtest ;+ ; NAME: ; sxt_vignette ; PURPOSE: ; Return the SXT X-ray vignette function determined from WSMR1 data. ; ; Called by sxt_off_axis to correct image shape (portion) given in index. ; CALLING SEQUENCE: ; vignette = sxt_vignette(index) ; Index must be a scalar structure. ; vignette = sxt_vignette(array) ; vignette = sxt_vignette(index,energy) ; ; INPUT: ; INDEX - Standard SXT index structure. Specifies portion of SXT FOV to return. ; Determine low or high energy from filter (if ENERGY argument not present). ; ARRAY - Vector of length 5 where: ; array(0) = x CCD coordinate & array(1) = y CCD coordinate ; array(2) = # of x pixels & array(3) = # of y pixels ; array(4) = Resolution as: 0=full, 1=half, 2=quarter ; OPTIONAL INPUT: ; ENERGY - If = 0 or not present, determine from index (filtb=4 or 5, energy=2) ; If = 1, use low-energy result (overrides index) ; If = 2, use high-energy result (overrides index) ; OUTPUT: ; 2-d matrix describing the SXT vignette function normalized to 1.0 at the ; position of highest effective area (gt_sxt_axis). ; PROCEDURE: ; The SXT vignette function consists of two non-concentric cones whose slopes ; and origins were determined from WSMR1 data and refined by analyis of flight ; data. See Calibration Notes 36 and 37. The origins are defined as: ; x y ; Inner-region (r < 21 arc-min) 515 633 (determined fr WSMR data) ; Outer-region (r > 21 arc-min) 533 599 (determined fr WSMR data) ; ; Note that the out-region centroid location and the cutoff radius (=21) may ; be changed in the future based on more extensive analysis of flight data. ; ; The slopes were derived from off-axis measurements at ; Low energy Al-K (8.3 A) and C-K (44 A) ; High energy Ag L (4.16 A) ; ; Be patient. On a DECstations 5000/200 this routine takes about ; 8 sec for a 512x 512 image ; 53 sec for a 1024x1024 image ; RESTRICTIONS: ; If index is passed in, must be a scalar ; MODIFICATION HISTORY: ; 12-Jan-94, J. R. Lemen LPARL, Written ; 3-Feb-94, JRL, Check that index is a scalar structure ; 4-Feb-94, JRL, Changed the outer-region cone centroid. ; 4-Oct-95, JRL, Fixed a bug that effected half-res and full-res PFIs ;- if N_elements(index) eq 0 then begin ; No input -- show the doc header doc_library,'sxt_vignette' return,-1 endif ; Determine the type of the input variable sz = size(index) typ = sz(1+sz(0)) ; NOTE: For half & quarter resolution, axis coordinates are divided by 2 ; and 4, respectively. This is slightly incorrect. wingaxis = [533.,599.] ; Axis of outer-region (derived from WSMR data) ; Axis of inner-region derived from gt_sxt_axis if (typ EQ 8) then begin ; Yes - then index = SXT index structure if n_elements(index) gt 1 then begin tbeep message,'Warning: index is NOT a scalar',/cont help,index message,'Will return vignette function for index(0)',/cont endif mode = gt_res(index(0)) ; Get resolution from "index" maxaxis = gt_sxt_axis(index(0)) ; Get the maximum effective axis maxaxis = maxaxis / (2^mode) ;*** Kluge for gt_sxt_axis wingaxis= wingaxis/ (2^mode) filtb = gt_filtb(index(0)) ; Get the filter if (filtb eq 4) or (filtb eq 5) then ienergy = 2 else ienergy = 1 x0 = gt_corner(index(0),/x) y0 = gt_corner(index(0),/y) dx = gt_shape(index(0),/x) ; Number of X pixels dy = gt_shape(index(0),/y) ; Number of Y pixels endif else begin if n_elements(index) eq 5 then mode = index(4) else begin message,'INDEX must be an SXT index or a vector of length 5',/cont return, -1 endelse maxaxis = gt_sxt_axis(0) ; Assume full resolution maxaxis = maxaxis / (2^mode) ;*** Kluge for gt_sxt_axis wingaxis= wingaxis/ (2^mode) ienergy = 1 ; Default to low-energy case x0 = index(0) & y0 = index(1) dx = index(2) & dy = index(3) ; The next statement allows us to produce plots for display showing the ; the knee at 21 arc-min if keyword_set(qtest) then wingaxis = maxaxis endelse x0 = x0 / (2^mode) ;*** Another Kluge/approx (JRL: 4-Oct-95) y0 = y0 / (2^mode) ;*** Another Kluge/approx (JRL: 4-Oct-95) ; Which energy should we use? if n_elements(energy) ne 0 then eenergy = energy else eenergy = 0 if (eenergy eq 0) then eenergy = ienergy ; Establish the slope and intercept parameters Rcut = 21.0 ; 21 arc-min (where the two cones intersect) if eenergy eq 1 then begin ; Low energy case slopein = -0.01244 ; relative intensity / arc-min slopeout = -0.04274 ; relative intensity / arc-min endif else begin ; High-energy case slopein = -0.02335 ; relative intensity / arc-min slopeout = -0.03328 ; relative intensity / arc-min endelse ; Set up the intercepts. Solve for outer-region cone at x = Rcut interceptin = 1.0 ; Y intercept for inner-region interceptout = interceptin + slopein * Rcut - slopeout * Rcut ; Convert slopes to per pixel [Full, half or quarter]: slopein = slopein / 60. * gt_pix_size(mode) ; rel int/pixel slopeout = slopeout / 60. * gt_pix_size(mode) ; rel int/pixel Rcut = Rcut * 60. / gt_pix_size(mode) ; = 513.7 pixels for full res ; - - - - - This table gives the values for Full-Resolution - - - - - - ; per pixel per arc-min ; Interceptin= 1.0 Both Energies ; ; slopein = -0.0005085 0.01244 Low Energy ; slopeout = -0.0017472 0.04274 ; Interceptout= 1.63630 Rcut = 21 arc-min ; ; slopein = -0.0009545 0.02335 High Energy ; slopeout = -0.0013605 0.03328 ; Interceptout= 1.20853 Rcut = 21 arc-min ; - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - ; Setup output variable cmask = fltarr(dx,dy) xp = maxaxis(0) & yp = maxaxis(1) ; Axis of inner-region xw = wingaxis(0) & yw = wingaxis(1) ; Axis of outer-region ;-------------------------------------------------------- ; Do the Hard work - Use a 1-d loop ; (This is faster than vectorizing, because uses less memory) xx = x0 + findgen(dx) ; X vector for j=y0,y0+dy-1 do begin ; Loop on Y rp = sqrt((xx-xp)^2 + (float(j)-yp)^2) peak = interceptin + slopein * rp rw = sqrt((xx-xw)^2 + (float(j)-yw)^2) wing = interceptout + slopeout * rw cmask(0,j-y0) = peak < wing ; Take the smaller of the two endfor ;-------------------------------------------------------- return, cmask end