imalign problem

Sinan Alis wrote on Nov 09, 2006

Hello,

I have some problems with imalign. When I use, it gave me error about coarse centering. I have been read through several old posts in the forum and didnt find exact solution to my problem. Actually it is related with my lack of knowledge. If I can get some info about boxsize and bigbox parameters, it seems to be solved.

OK, I have 160 images to be aligned. And because of tracking problems there are a 170 pixels in x and 280 pixels in y difference at most between the extreme images. So what should be my boxsize and bigbox parameters? Also tolerance.

Here is my error. This error came out with standart imalign parameters (below).

*******************************************************************
im> imalign
Input images (@im.lis):
Reference image (v080.fit):
Reference coordinates file (im.coords):
Output images (@ima.lis):
#Coords Image X-center Err Y-center Err Num
v001.fit 832.147 (0.465) 617.443 (0.540) 1
Warning: failed to converge near (484,563)
v001.fit 507.701 (0.473) 409.049 (0.459) 3
v001.fit 805.730 (0.374) 516.214 (0.393) 4
v001.fit 741.918 (0.461) 227.159 (0.337) 5

v002.fit 831.789 (0.430) 617.256 (0.515) 1
v002.fit 484.189 (0.531) 563.378 (0.545) 2
Warning: failed to converge near (508,407)
v002.fit 809.203 (0.412) 516.842 (0.387) 4
v002.fit 743.060 (0.476) 226.150 (0.487) 5

Warning: failed to converge near (832,617)
Warning: Problem with coarse centering
ERROR: No shifts were calculated.
*************************************************************************

Here is my imalign parameters

*************************************************************************
im> lpar imalign
input = "@im.lis" Input images
reference = "v080.fit" Reference image
coords = "im.coords" Reference coordinates file
output = "@ima.lis" Output images
(shifts = "") Initial shifts file
(boxsize = 7) Size of the small centering box
(bigbox = 11) Size of the big centering box
(negative = no) Are the features negative ?
(background = INDEF) Reference background level
(lower = INDEF) Lower threshold for data
(upper = INDEF) Upper threshold for data
(niterate = 3) Maximum number of iterations
(tolerance = 0) Tolerance for convergence
(maxshift = INDEF) Maximum acceptable pixel shift
(shiftimages = yes) Shift the images ?
(interp_type = "linear") Interpolant
(boundary_typ = "nearest") Boundary type
(constant = 0.) Constant for constant boundary extension
(trimimages = yes) Trim the shifted images ?
(verbose = yes) Print the centers, shifts, and trim section ?
(list = "")
(mode = "ql")
********************************************************************************

Additional info: When I set the tolerance to 3, it gives no error but nothing happens to images.

Thanks a lot...

Sinan Alis

Istanbul University
Astronomy and Space Sciences Dept.
34119, University - Istanbul / TURKEY
salis@istanbul.edu.tr

---

Mike Fitzpatrick wrote on Nov 09, 2006

I think the problem is that you don't specify any 'shift' parameter. The task assumes your coordinate list are the (approximate) center of objects in the reference image. For each image in the list the corresponding line from the shifts file is used to offset these reference coords to find the same star in the shifted image, the centroiding is then done using a box of the size given by the boxsize/bigbox params to compute the final shift for alignment. With no 'shifts' file the reference coords are used on each image in the list, and since you say there is an extreme of (170,280) pixel shift between images the chances are there won't be a star at the same position in all images and so you get the message.

Try specifying an initial shift for each image, and set the box params to be about the size of a star in the field. See also the WREGISTER/SREGISTER task for ways to align the image by the WCS, or XREGISTER by cross correlation. Hope this helps.

Cheers,
-Mike

---

Last post on Nov 09, 2006