[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: What Is It

What spots are you referring to?  I see a couple of cosmic rays, and three splotches.
The middle one looks like frost to me, and so I would suspect the others to be frost
as well.

Traps look like cosmic rays that are trailed along the column direction.  You can't
miss them.  They are most obvious in low light level exposures, and with your high
dark current, you may not see them.  (One method of removing them from science exposures
was to add a 'fat zero', and that is what dark current does automatically.)

Now on to your other questions.  (I didn't go into the office over the weekend like I
normally do.  Too many outside projects at home that need to be taken care of before
winter sets in!)

The noise should not increase linearly since it is a square root function.  If you
examine it on a log-linear plot, then it will be linear with a slope of 1/2.  If I
understand your mean DN level, you had more than 32768 DN in each pixel, which seems
abnormally high -- did you have the bias set properly?  Mike: does Image Scientist
use double-precision floating point when calculating means and standard deviations?

I've used a spectrum analyzer for hunting down noise sources.  Knowing the frequency
and the intensity of the various components helps.  I think there are some cheap ones,
some that plug into PC busses and/or have PC software.  Our requirements are more
stringent than TASS since we have no dark current and at times either use very short
exposures or narrow band filters or spectroscopy, where sky noise does *not* dominate.

Once you get down into the few e- noise level, cosmic rays do affect the standard
deviation calculation.  We either remove them before the calculation, or use a noise
histogram and take the modal value.  When you can start taking flats, you might
try the 2flat/2dark scheme to determine noise and gain to ensure that your calculations
are correct.