problems with datamax,datamin in ccfind task.
Tiina Liimets wrote on Nov 20, 2008
Hello,
I am doing astrometry to a bunch of images from a different telescopes. I have gotten help from here already once ( "CCMAP output as an input for GEOTRAN?", posted: Tue Jul 15, 2008 1:33 pm) but now I am having another problem.
The parameters datamax and datamin in ccfind task does not seem to work as I understand from the help. From the help file I get: "datamin = INDEF, datamax = INDEF. The minimum and maximum good data values. Values outside this range are exclude from the x and y marginal centroid computation.
Bad data can be excluded from the centroiding algorithm by setting the datamin and datamax parameters"
Isnt it suppose to mean the "bad" stars are rejected? My experience says that it does not work like that.
I put the datamax=50000 (the saturation level is 65000) but it is still finding the stars which have highest peak 68000. CCMAP then rejects some of the overexposed stars but not all. I still have some saturated stars contributing to the plate solution. Setting datamax=25000 makes sure that none of the saturated stars are chosen for plate solution (CCMAP) but some of them are still found by CCFIND (highest peak ~65000). How does parameter datamax influence CCMAP?
I would like to remove the bad stars (too faint, overexposed) in CCFIND task, before starting ccmap to be sure that the "bad" stars are not included to the plate solution calculations. Is that possible in CCFIND?
For datamin I use a value which is 10 times higher than the stdev of the background and I find the similar problem: the value for datamin does not reflect the rejected stars highest peak. In one example it rejects the stars which have 30 times higher peak value than giving by datamin.
NB! My images are sky subtracted.
I add here a parts of my script for making astrometry:
...
ccfind (input="in.dat", output=imageinp//"_wpix.match", images=imageinp, xref=x_pix, yref=y_pix, xmag=scaleinp, ymag=scaleinp, xrotati=180, yrotati=0, lngref=7:4:4.819, latref=-3:50:50.69, sbox=20, cbox=9, datamin=stdev10, datamax=25000, >>& imageinp//"_logfile.txt")
ccmap (input=imageinp//"_wpix.match", database= imageinp//"_wpix.db", lngcol=1, latcol=2, xcol=3, ycol=4, refpoint="user", lngref=7:4:4.819, latref=-3:50:50.69, results=imageinp//"_ccmap_results", refsystem="j2000", projection="tnx", fitgeom="general", functio="polynomial", xxorder=3, xyorder=3, yxorder=3, yyorder=3, xxterms="half", yxterms="half", maxiter=3, reject=3, interact+, >>& imageinp//"_logfile.txt")
ccsetwcs (images=imageinp, database=imageinp//"_wpix.db", solution=imageinp//"_wpix.match", update+, >>& imageinp//"_logfile.txt")
...
All the best,
Tiina
I am doing astrometry to a bunch of images from a different telescopes. I have gotten help from here already once ( "CCMAP output as an input for GEOTRAN?", posted: Tue Jul 15, 2008 1:33 pm) but now I am having another problem.
The parameters datamax and datamin in ccfind task does not seem to work as I understand from the help. From the help file I get: "datamin = INDEF, datamax = INDEF. The minimum and maximum good data values. Values outside this range are exclude from the x and y marginal centroid computation.
Bad data can be excluded from the centroiding algorithm by setting the datamin and datamax parameters"
Isnt it suppose to mean the "bad" stars are rejected? My experience says that it does not work like that.
I put the datamax=50000 (the saturation level is 65000) but it is still finding the stars which have highest peak 68000. CCMAP then rejects some of the overexposed stars but not all. I still have some saturated stars contributing to the plate solution. Setting datamax=25000 makes sure that none of the saturated stars are chosen for plate solution (CCMAP) but some of them are still found by CCFIND (highest peak ~65000). How does parameter datamax influence CCMAP?
I would like to remove the bad stars (too faint, overexposed) in CCFIND task, before starting ccmap to be sure that the "bad" stars are not included to the plate solution calculations. Is that possible in CCFIND?
For datamin I use a value which is 10 times higher than the stdev of the background and I find the similar problem: the value for datamin does not reflect the rejected stars highest peak. In one example it rejects the stars which have 30 times higher peak value than giving by datamin.
NB! My images are sky subtracted.
I add here a parts of my script for making astrometry:
...
ccfind (input="in.dat", output=imageinp//"_wpix.match", images=imageinp, xref=x_pix, yref=y_pix, xmag=scaleinp, ymag=scaleinp, xrotati=180, yrotati=0, lngref=7:4:4.819, latref=-3:50:50.69, sbox=20, cbox=9, datamin=stdev10, datamax=25000, >>& imageinp//"_logfile.txt")
ccmap (input=imageinp//"_wpix.match", database= imageinp//"_wpix.db", lngcol=1, latcol=2, xcol=3, ycol=4, refpoint="user", lngref=7:4:4.819, latref=-3:50:50.69, results=imageinp//"_ccmap_results", refsystem="j2000", projection="tnx", fitgeom="general", functio="polynomial", xxorder=3, xyorder=3, yxorder=3, yyorder=3, xxterms="half", yxterms="half", maxiter=3, reject=3, interact+, >>& imageinp//"_logfile.txt")
ccsetwcs (images=imageinp, database=imageinp//"_wpix.db", solution=imageinp//"_wpix.match", update+, >>& imageinp//"_logfile.txt")
...
All the best,
Tiina
Francisco Valdes wrote on Nov 20, 2008
Hello Tina,
The datamin/datamax parameters exclude "pixels" from a centroid fit and not whole stars. So as long as the number of pixels excluded by the datamin/datamax do not cause the centroid determination to fail you will still get a coordinate for saturated or bright stars.
Yes it is good to select only bright unsaturated stars but CCFIND does not do that for you. The implied approach is that the input list of coordinates has been filtered before giving it to the task. For example, if you get the reference coordinates from some reference catalog (say USNO-B1 or 2Mass) you could include a magnitude cut to eliminate bright and faint sources (where this depends on the characteristics of your exposure).
The other common thing to do is to not be picky about the stars going into ccmap but then when you do the fit interactively kill the obvious outliers or non-interactively use sigma clipping to get rid of outliers. More significant than the slight increase in centroid errors from faint and bright stars is usually the mismatch in epochs of the image and reference coordinates; i.e. not correcting for proper motions.
Yours,
Frank Valdes
The datamin/datamax parameters exclude "pixels" from a centroid fit and not whole stars. So as long as the number of pixels excluded by the datamin/datamax do not cause the centroid determination to fail you will still get a coordinate for saturated or bright stars.
Yes it is good to select only bright unsaturated stars but CCFIND does not do that for you. The implied approach is that the input list of coordinates has been filtered before giving it to the task. For example, if you get the reference coordinates from some reference catalog (say USNO-B1 or 2Mass) you could include a magnitude cut to eliminate bright and faint sources (where this depends on the characteristics of your exposure).
The other common thing to do is to not be picky about the stars going into ccmap but then when you do the fit interactively kill the obvious outliers or non-interactively use sigma clipping to get rid of outliers. More significant than the slight increase in centroid errors from faint and bright stars is usually the mismatch in epochs of the image and reference coordinates; i.e. not correcting for proper motions.
Yours,
Frank Valdes
Tiina Liimets wrote on Nov 20, 2008
Hello Frank,
thank you for the reply.
I wanted to make the astrometry script uniform to use for many different size and deepness images. I have around 400 frames and my idea was to write a task which is usable for all of them. It would take too much time to sort out how deep is every images and then choose the input coordinate list accordingly.
"The other common thing to do is to not be picky about the stars going into ccmap but then when you do the fit interactively kill the obvious outliers or non-interactively use sigma clipping to get rid of outliers."
yes, that is what I am doing now. Many times the sigma clipping works well enough and I only have to correct for couple of stars. Sometimes more time consuming correction is needed but at least I get out the "bad" stars.
"More significant than the slight increase in centroid errors from faint and bright stars is usually the mismatch in epochs of the image and reference coordinates; i.e. not correcting for proper motions. "
yes. I have been warned about that.
Regards,
Tiina
thank you for the reply.
I wanted to make the astrometry script uniform to use for many different size and deepness images. I have around 400 frames and my idea was to write a task which is usable for all of them. It would take too much time to sort out how deep is every images and then choose the input coordinate list accordingly.
"The other common thing to do is to not be picky about the stars going into ccmap but then when you do the fit interactively kill the obvious outliers or non-interactively use sigma clipping to get rid of outliers."
yes, that is what I am doing now. Many times the sigma clipping works well enough and I only have to correct for couple of stars. Sometimes more time consuming correction is needed but at least I get out the "bad" stars.
"More significant than the slight increase in centroid errors from faint and bright stars is usually the mismatch in epochs of the image and reference coordinates; i.e. not correcting for proper motions. "
yes. I have been warned about that.
Regards,
Tiina
Last post on Nov 20, 2008