Help with geomap/geotran
mconnelley wrote on May 09, 2008
Hello:
I'm trying to get several images aligned with each other that were taken with a variety of cameras with a range of plate scales. For the time being, I'm just trying to do this one image pair at a time. Geomap seems to work ok, in the sense that the xscale and yscale and rotations look about right. However, when I run geotran, the image is much smaller when it is supposed to be larger by about a factor of 4. According to the database file made by gemap, xmag and ymag are both around 4, which is what they should be. The database file produced by geomap (1990_10_05.database.txt) is:
# Fri 16:24:04 09-May-2008
begin 1990_10_05.input.txt
xrefmean 337.684
yrefmean 159.392
xmean 1346.98
ymean 723.266
geometry general
function polynomial
xshift -20.31994
yshift 75.23201
xmag 4.052993
ymag 4.06612
xrotation 0.002998151
yrotation 359.8823
xrms 0.2887288
yrms 0.1302827
surface1 11
3. 3.
2. 2.
2. 2.
0. 0.
1. 1.
3000. 3000.
1. 1.
3000. 3000.
-20.31994 75.23201
4.052993 -2.120834E-4
-0.008351182 4.066111
surface2 0
The input file (1990_10_05.input.txt) to geomap is:
320.84 196.03 1278.58 872.07
312.73 92.26 1246.25 450.24
409.70 174.15 1639.09 783.24
401.30 213.45 1603.97 943.19
243.85 121.07 967.01 567.59
Here, columns 1 and 2 are the x and y coordinates of 5 stars in the image that I want to warp, and columns 3 and 4 are the coordinates of the stars in my reference image (which I want all other images to match). The commands I used are:
geomap 1990_10_05.input.txt 1990_10_05.database.txt 1. 3000. 1. 3000.
geotran L483_KP_1990_10_05.fits L483_KP_1990_10_05_warped.fits 1990_10_05.database.txt 1990_10_05.input.txt
The reference image is rather large, so I used 1 to 3000 for the minimum and maximum values. According to DS9, the image pixel coordinates go from (88,98) to (2752, 1329). I believe that this image is a mosaic made from three smaller images, which explains the odd image size.
As I mentioned above, geotran makes the image smaller by a factor of 4.24 in X and 5.66 in Y instead of making the imager larger. I got the a similar result using gregister. Does anyone have an idea as to what is going on? I'll be happy to provide whatever information would be helpful.
Cheers
Mike Connelley
I'm trying to get several images aligned with each other that were taken with a variety of cameras with a range of plate scales. For the time being, I'm just trying to do this one image pair at a time. Geomap seems to work ok, in the sense that the xscale and yscale and rotations look about right. However, when I run geotran, the image is much smaller when it is supposed to be larger by about a factor of 4. According to the database file made by gemap, xmag and ymag are both around 4, which is what they should be. The database file produced by geomap (1990_10_05.database.txt) is:
# Fri 16:24:04 09-May-2008
begin 1990_10_05.input.txt
xrefmean 337.684
yrefmean 159.392
xmean 1346.98
ymean 723.266
geometry general
function polynomial
xshift -20.31994
yshift 75.23201
xmag 4.052993
ymag 4.06612
xrotation 0.002998151
yrotation 359.8823
xrms 0.2887288
yrms 0.1302827
surface1 11
3. 3.
2. 2.
2. 2.
0. 0.
1. 1.
3000. 3000.
1. 1.
3000. 3000.
-20.31994 75.23201
4.052993 -2.120834E-4
-0.008351182 4.066111
surface2 0
The input file (1990_10_05.input.txt) to geomap is:
320.84 196.03 1278.58 872.07
312.73 92.26 1246.25 450.24
409.70 174.15 1639.09 783.24
401.30 213.45 1603.97 943.19
243.85 121.07 967.01 567.59
Here, columns 1 and 2 are the x and y coordinates of 5 stars in the image that I want to warp, and columns 3 and 4 are the coordinates of the stars in my reference image (which I want all other images to match). The commands I used are:
geomap 1990_10_05.input.txt 1990_10_05.database.txt 1. 3000. 1. 3000.
geotran L483_KP_1990_10_05.fits L483_KP_1990_10_05_warped.fits 1990_10_05.database.txt 1990_10_05.input.txt
The reference image is rather large, so I used 1 to 3000 for the minimum and maximum values. According to DS9, the image pixel coordinates go from (88,98) to (2752, 1329). I believe that this image is a mosaic made from three smaller images, which explains the odd image size.
As I mentioned above, geotran makes the image smaller by a factor of 4.24 in X and 5.66 in Y instead of making the imager larger. I got the a similar result using gregister. Does anyone have an idea as to what is going on? I'll be happy to provide whatever information would be helpful.
Cheers
Mike Connelley
Mike Fitzpatrick wrote on May 09, 2008
I can't tell what's happening based on your post and would ask that you put up a URL to the image/db files so I can reproduce it. You might also look at the WREGISTER task as a way to use to WCS to register the images, this would take into account the terms that represent the cutout where DS9 shows the image coords don't start at (0,0) or are similar sizes.
Cheers,
-Mike
Cheers,
-Mike
Rob Steele wrote on May 09, 2008
Curious if you ever got this working.
Last post on May 09, 2008