Recognising blank pixels in FITS images

[reynolds@…]

I've been working with some FITS images that have a large number of blank pixels. The images were created with AIPS and the blank pixels have 'nan' as a value, rather than the -1 that Duchamp seems to require, and as a result the statistics get a bit messed up (the flux threshold always comes out as -8.00061). Would it be possible to have Duchamp recognise values other than -1 to mean the pixel is blanked, perhaps with this as a configurable parameter?

BTW, I find Duchamp to be much the best source finding software that I have used - nice work.

Thanks, Cormac Reynolds.

[MatthewWhiting]

Hi Cormac,

Thanks for your feedback! Sorry it's taken a little while to get back to you -- there were some issues with our trac system not sending out notifications of tickets, so I only saw this recently.

I actually used to have a "blankPixValue" parameter that the user could set the value with, but I got rid of it in favour of just reading the blank pixel value from the FITS header. That is why the value of -8.00061 appears -- I had that set as the default for no reason other than that was the value in a bunch of data I was using for testing.

I could easily reinstate the blankPixValue parameter, but that probably wouldn't help in the case of 'nan' as the value. I'll have a look at how cfitsio flags blank pixels, and get back to you.

Cheers, Matthew Whiting.

[MatthewWhiting]

Hi Cormac,

I'm revisiting old Duchamp tickets in an effort to clear things up before the next release. I've cc'd this to your Curtin address -- hope that works.

Does this problem with the blank pixels still apply? If so, would you be able to point me to a file that causes the problem?

Otherwise I'll close the ticket.

[MatthewWhiting]

No reply from Cormac, so I'm closing the ticket.