crosscor finds zero shift for shifted images

kanaan wrote on Nov 19, 2007

Hi folks,

I have used crosscor (stsdas.analysis.fourier) for ages in my high speed photometry package to track moving fields of stars. I only use it when using the previous position or immatch (from ctio) fail. Crosscor is a lot slower, but it never fails, well, hadn't failed yet.

I have data from one particular observatory where the crosscorrelation looks really funny. Whatever I try, no matter how shifted two images are, the cross correlation always comes back with a cross centered at the crosscorrelation image center. Even copying an image shifting it and then doing cross correlation returns the same zero shift result.

I have tapered the image (as I always do), I have removed cosmic rays from the images themselves as well as from the crosscorrelation image. ALways the same result. I have also looked for bad pixels, bad columns, tried different image sections...

If anyone has some time to take a look at this I would appreciate a suggestion.

I have put a couple images from this site at:

www.astro.ufsc.br/~kanaan/tmp/cc.tar.bz2

in this tar file:

532800 lixo/LCO00275.fit problem image
532800 lixo/LCO00265.fit problem image
1056960 lixo/lixo16.fits cross-correlation between lixo1tcr and lixo2tcr
532800 lixo/LCO00275cr.fits cosmic ray removed
532800 lixo/LCO00265cr.fits cosmic ray removed
532800 lixo/lixo1t.fits LCO00265 tapered 35% edge cosine bell
532800 lixo/lixo2t.fits LCO00275 tapered 35% edge cosine bell
532800 lixo/lixo1tcr.fits LCO00265cr tapered
532800 lixo/lixo2tcr.fits LCO00275cr tapered

Antonio

Antonio Kanaan
Departamento de Fisica - Universidade Federal de Santa Catarina

---

Francisco Valdes wrote on Nov 19, 2007

Hello Antonio,

In looking at posts with no reply I see you did not get any help. I tried to get your data but, reasonably, it was gone. If you want someone to also look at the problem you can let me know.

Yours,
Frank Valdes

---

Last post on Nov 19, 2007