PDA

View Full Version : sensor/system calibration: dark = bias + thermal


January 7th 04, 07:20 PM
I read(*) that a standard practice is to collect a set of "bias"
frames and average them together. You then collect another set of
"dark" frames (at some fixed exposure), remove the aforecollected
bias, averge them up and you obtain a "thermal" frame which can be
scaled by exposure time.

My question: is there a need to collect the bias frames at all? If
instead you collected a set of "dark" frames at a number of exposure
times (say 1s, 2s, 4s, 8s, ..., 32s, 64s, etc), could you not extract
the "bias" information from them, as well as obtaining the thermal
scaling? If so, then you not only save the time needed to collect the
bias frames, store and process them, but you can also measure a
goodness-of-fit to the assumed linear model that "dark = bias +
thermal*exposure".

Or is there some Mystery to this process that makes the extra time and
space for separate bias frames desirable?

(*) various from www.google.com, as well as Berry/Burnell "Handbook of
Astronomical Image Processing", 2000.

Mark Ritchie
January 8th 04, 04:20 PM
Hi,

My take on this is that the bias frame is used to get the noise that is
constant (noise caused by the actual readout of the CCD data), and therefore
does not change with exposure setting or temperature. I think you should be
able to create one master bias frame (a zero second exposure) and use it
with all of your exposures.

Now, the thermal frame gets the noise that varies with time and temperature,
and I guess is assumed to be linear (otherwise you couldn't scale it). I
think you should take a set of dark frames for each imaging session so that
you get the same temperature as the real exposure, but if you end up using
the bias frame, then you don't have to have a dark frame of exactly the same
exposure as your real frames. You would subtract the bias frame from the
dark frame, and then scale the resulting thermal frame. You then subtract
both the bias frame and the scaled thermal frame.

I just take a set of dark frames of the same exposure time as my real
images, and then just subtract them without any scaling, but then again, I
am a rank beginner using a SAC8 (modified security cam).

Mark Ritchie

> wrote in message
om...
> I read(*) that a standard practice is to collect a set of "bias"
> frames and average them together. You then collect another set of
> "dark" frames (at some fixed exposure), remove the aforecollected
> bias, averge them up and you obtain a "thermal" frame which can be
> scaled by exposure time.
>
> My question: is there a need to collect the bias frames at all? If
> instead you collected a set of "dark" frames at a number of exposure
> times (say 1s, 2s, 4s, 8s, ..., 32s, 64s, etc), could you not extract
> the "bias" information from them, as well as obtaining the thermal
> scaling? If so, then you not only save the time needed to collect the
> bias frames, store and process them, but you can also measure a
> goodness-of-fit to the assumed linear model that "dark = bias +
> thermal*exposure".
>
> Or is there some Mystery to this process that makes the extra time and
> space for separate bias frames desirable?
>
> (*) various from www.google.com, as well as Berry/Burnell "Handbook of
> Astronomical Image Processing", 2000.

Richard Crisp
January 9th 04, 04:33 AM
"Mark Ritchie" > wrote in message
...
> Hi,
>
> My take on this is that the bias frame is used to get the noise that is
> constant (noise caused by the actual readout of the CCD data), and
therefore
> does not change with exposure setting or temperature. I think you should
be
> able to create one master bias frame (a zero second exposure) and use it
> with all of your exposures.


Actually it is the bias that varies with time. In fact bias can vary by a
few ADU counts over a few hours.

See "The Handbook of Astronomical Image Processing" by Berry and Burnell
chapter 4 for more details


>
> Now, the thermal frame gets the noise that varies with time and
temperature,
> and I guess is assumed to be linear (otherwise you couldn't scale it). I
> think you should take a set of dark frames for each imaging session so
that
> you get the same temperature as the real exposure, but if you end up using

What you really want to do is to take a set of master darks and master dark
biases one day. You median or otherwise combine the biases to create a
single master. Ditto for the Dark. Then subtract the Master Bias from the
Master Dark and keep the bias subracted Dark as your library dark.

When you take images on another day, you simply shoot maybe 10 biases which
represent that bias on that particualar day. You combine them together and
add that to your Library Dark. Then you use that bias-added Library Dark for
your Calibration Dark.

Think about it: Taking 10 darks of , say, 30 minutes takes a lot of time. On
the other hand 10 biases are much faster: just download time.

There's little point in keeping library biases and taking darks each time.
The idea behind keeping a library of darks is to save time.

I find that so long as I am operating at the same temp, I can use
bias-subtracted library darks for months. I just shoot a set of biases each
imaging session.

Mark Ritchie
January 16th 04, 03:22 PM
Richard,
I just got "The Handbook of Astronomical Image Processing" book and
software for Christmas, and I did read Chapter 4, but obviously I didn't
understand it ! I re-read the chapter, and I now see that I had it pretty
much backwards. I guess I was reading it with a 'biased' opinion of how the
bias frames work.

I am taking such short exposures (30 seconds max), that I have never done
anything more complicated than just taking darks that match my real images,
so I have never tried to actually use a bias frame, which I guess should
have made me think twice about responding to the original post in the first
place.

Anyway, thanks for correcting me on this, I certainly try not to spread bad
information.

Mark Ritchie

"Richard Crisp" > wrote in message
m...
>
> "Mark Ritchie" > wrote in message
> ...
> > Hi,
> >
> > My take on this is that the bias frame is used to get the noise that is
> > constant (noise caused by the actual readout of the CCD data), and
> therefore
> > does not change with exposure setting or temperature. I think you should
> be
> > able to create one master bias frame (a zero second exposure) and use it
> > with all of your exposures.
>
>
> Actually it is the bias that varies with time. In fact bias can vary by a
> few ADU counts over a few hours.
>
> See "The Handbook of Astronomical Image Processing" by Berry and Burnell
> chapter 4 for more details
>
>
> >
> > Now, the thermal frame gets the noise that varies with time and
> temperature,
> > and I guess is assumed to be linear (otherwise you couldn't scale it). I
> > think you should take a set of dark frames for each imaging session so
> that
> > you get the same temperature as the real exposure, but if you end up
using
>
> What you really want to do is to take a set of master darks and master
dark
> biases one day. You median or otherwise combine the biases to create a
> single master. Ditto for the Dark. Then subtract the Master Bias from the
> Master Dark and keep the bias subracted Dark as your library dark.
>
> When you take images on another day, you simply shoot maybe 10 biases
which
> represent that bias on that particualar day. You combine them together and
> add that to your Library Dark. Then you use that bias-added Library Dark
for
> your Calibration Dark.
>
> Think about it: Taking 10 darks of , say, 30 minutes takes a lot of time.
On
> the other hand 10 biases are much faster: just download time.
>
> There's little point in keeping library biases and taking darks each time.
> The idea behind keeping a library of darks is to save time.
>
> I find that so long as I am operating at the same temp, I can use
> bias-subtracted library darks for months. I just shoot a set of biases
each
> imaging session.
>
>
>

Richard Crisp
February 16th 04, 12:53 PM
I am happy I could help Mark.

Tell me have you teamed up with any one named Chretien yet?

(yes I know that it needs to be spelled Ritchey, but I couldn't resist).


"Mark Ritchie" > wrote in message
...
> Richard,
> I just got "The Handbook of Astronomical Image Processing" book and
> software for Christmas, and I did read Chapter 4, but obviously I didn't
> understand it ! I re-read the chapter, and I now see that I had it pretty
> much backwards. I guess I was reading it with a 'biased' opinion of how
the
> bias frames work.
>
> I am taking such short exposures (30 seconds max), that I have never done
> anything more complicated than just taking darks that match my real
images,
> so I have never tried to actually use a bias frame, which I guess should
> have made me think twice about responding to the original post in the
first
> place.
>
> Anyway, thanks for correcting me on this, I certainly try not to spread
bad
> information.
>
> Mark Ritchie
>
> "Richard Crisp" > wrote in message
> m...
> >
> > "Mark Ritchie" > wrote in message
> > ...
> > > Hi,
> > >
> > > My take on this is that the bias frame is used to get the noise that
is
> > > constant (noise caused by the actual readout of the CCD data), and
> > therefore
> > > does not change with exposure setting or temperature. I think you
should
> > be
> > > able to create one master bias frame (a zero second exposure) and use
it
> > > with all of your exposures.
> >
> >
> > Actually it is the bias that varies with time. In fact bias can vary by
a
> > few ADU counts over a few hours.
> >
> > See "The Handbook of Astronomical Image Processing" by Berry and Burnell
> > chapter 4 for more details
> >
> >
> > >
> > > Now, the thermal frame gets the noise that varies with time and
> > temperature,
> > > and I guess is assumed to be linear (otherwise you couldn't scale it).
I
> > > think you should take a set of dark frames for each imaging session so
> > that
> > > you get the same temperature as the real exposure, but if you end up
> using
> >
> > What you really want to do is to take a set of master darks and master
> dark
> > biases one day. You median or otherwise combine the biases to create a
> > single master. Ditto for the Dark. Then subtract the Master Bias from
the
> > Master Dark and keep the bias subracted Dark as your library dark.
> >
> > When you take images on another day, you simply shoot maybe 10 biases
> which
> > represent that bias on that particualar day. You combine them together
and
> > add that to your Library Dark. Then you use that bias-added Library Dark
> for
> > your Calibration Dark.
> >
> > Think about it: Taking 10 darks of , say, 30 minutes takes a lot of
time.
> On
> > the other hand 10 biases are much faster: just download time.
> >
> > There's little point in keeping library biases and taking darks each
time.
> > The idea behind keeping a library of darks is to save time.
> >
> > I find that so long as I am operating at the same temp, I can use
> > bias-subtracted library darks for months. I just shoot a set of biases
> each
> > imaging session.
> >
> >
> >
>
>