affordable camera suggestions

[Peter van Loon]

Dear All,

I am looking for a camera compatible with ImageJ.
It should have the next specs: at least 2-3 Mega Pixel, colour and C-mount
Surfing the internet I see only expensive solutions. Does anyone know some
affordable solutions (2000-3000 euro max),

Thanks in advance!,

Peter van Loon

The information contained in this e-mail (and attachments if any) is exclusively intended for the recipient(s) named above and may be confidential, proprietary, and/or legally privileged. Unintentional disclosure of this e-mail does not constitute a waiver of any right or privilege. Please notify us if you are not the intended recipient and do not use, print, copy, forward, or disclose (any part of) this e-mail, but delete it (and attachments and copies if any) subsequently. Thank you.

You can find disclaimer translations in different languages by visiting our website http://www.rijkzwaan.com/edisclaimer

The registration number of each Rijk Zwaan company can be found by using the following link:
http://www.rijkzwaan.com/registrationnumbers

[bradscopegems-2]

If your requirement is for still images, even a medium-price digital SLR camera such as the Nikon D90 will give superb results at higher resolution (14 Megapixels). You will also have to buy a an adaptor to attach such a camera to a c-mount. These cost around 400 euros from a company in Austria, see http://www.lmscope.com/produkt22/LM Mikroskop Adapter Mikroskope en.shtml  but also check Meiji Techno as a possible source, or Martin Microscopes if you are in USA.  There is advice on the Austrian website on which DSLR is best. High-end Canon Eos models seem to have the best computer control software, but may exceed your budget.
  If you are after video,  the Nikon D90 gives high-definition streaming video  (so-called 'Live View') which is compressed but looks superb on a HDMI monitor. This is very useful if you need to focus a microscope or other device. High-end DSLRs tend to provide the best high-resolution video  outputs.
   Very cheap USB cameras (around 400 euros)  which have a c-mount ( e.g. GXCam) are available which can give up to 5 Mpixel resolution in colour, but the refresh rate is only 5 per second, which is a pain for focussing and following movement. These come with everything necessary for attaching to a microscope.

[Gabriel Landini]

On Saturday 12 Feb 2011, bradscopegems wrote:
> If your requirement is for still images, even a medium-price digital SLR
> camera such as the Nikon D90 will give superb results at higher resolution
> (14 Megapixels).

The resolution is mainly driven by the microscope optics. With a 14 MP camera
one gets what is called "empty magnification".
I would suggest to find out what are the resolution values of the microscope
objectives are (from the manufacturer manuals) and start thinking from there.
Depending on the field width of the camera and the optics, I think anything >
4 MP or thereabouts will not capture much more detail than the resolution of
the optical system will be able to provide in a relatively good brightfield
microscope.

Of course one can subsample large images with empty magnification and reduce
file size while keeping the useful amount of detail, but there are other
disadvantages. With a SLR one will struggle to make illumination modifications
on the fly to correct for brightfield background illumination using the
transmittance method. You can do it, but it will be more time consuming as one
cannot check histogram saturation on the fly, etc. then transfer the images to
the computer, reload them, etc. Whether this is crucial it depends on how many
images one is expecting to acquire.

Also remember that colour cameras using Bayer masks will interpolate colours
and that reduces further the resolution of the image data. A good alternative
is to use greyscale cameras with a R-G-B filter wheel or tunable filter so
each pixel is exposed 3 times (R G & B). This avoids the interpolation. Such
things are, sadly more expensive.

If I had to buy a new camera, I would try to find out what is supported right
now (some camera manufacturers produce IJ plugins that can drive their
cameras, but be aware that some of these only support manual acquisition, and
not driven from a plugin or macro, which is really useful).
Check the MicroManager pages to see what it is supported too.
If you decide to get a camera supported by an ImageJ plugin but without macro
support, you could still use the IJ_Robot plugin to try to automate it. It is
an ugly way of doing it, but it works.

I hope this is useful.
Cheers

Gabriel

[bradscopegems-2]

Dear Gabriel,
                     You will not find resolution values in the scant
literature supplied by microscope manufacturers, but you can calculate how
many pixels you will need from the N.A. and magnification written on the
objective.  Assuming Nyquist sampling, the pixel separation (assuming no
extra magnification) is  21, 14 and 16 microns for three commonly-used
objectives (60x N.A. 1.4, 20x N.A. 0.7 and 10x N.A. 0.3). So, if we wish to
capture a 20 x 20 mm square area of the intermediate image (a reasonably
large fraction of what we see in an eyepiece)  we need 0.9, 2 and 1.5
megapixels minimum (i.e. to record full detail without any empty
magnification. In practice, microscopists usually choose to work with about
3x the linear magnification at the Nyquist minimum,  so this means that the
preferred number of pixels would be nine times this, unless the camera was
recording a reduced area of the intermediate image. This means that the high
pixel numbers of the modern  DSLRs are not overkill.  This is why standard
PAL video resolution (about 0.4 megapixels maximum) is so hopeless for
microscopy, forcing users to capture only a tiny fraction of the eyepiece
field.
Brad Amos

On 12 February 2011 12:20, Gabriel Landini <[hidden email]> wrote:

> On Saturday 12 Feb 2011, bradscopegems wrote:
> > If your requirement is for still images, even a medium-price digital SLR
> > camera such as the Nikon D90 will give superb results at higher
> resolution
> > (14 Megapixels).
>
> The resolution is mainly driven by the microscope optics. With a 14 MP
> camera
> one gets what is called "empty magnification".
> I would suggest to find out what are the resolution values of the
> microscope
> objectives are (from the manufacturer manuals) and start thinking from
> there.
> Depending on the field width of the camera and the optics, I think anything
> >
> 4 MP or thereabouts will not capture much more detail than the resolution
> of
> the optical system will be able to provide in a relatively good brightfield
> microscope.
>
> Of course one can subsample large images with empty magnification and
> reduce
> file size while keeping the useful amount of detail, but there are other
> disadvantages. With a SLR one will struggle to make illumination
> modifications
> on the fly to correct for brightfield background illumination using the
> transmittance method. You can do it, but it will be more time consuming as
> one
> cannot check histogram saturation on the fly, etc. then transfer the images
> to
> the computer, reload them, etc. Whether this is crucial it depends on how
> many
> images one is expecting to acquire.
>
> Also remember that colour cameras using Bayer masks will interpolate
> colours
> and that reduces further the resolution of the image data. A good
> alternative
> is to use greyscale cameras with a R-G-B filter wheel or tunable filter so
> each pixel is exposed 3 times (R G & B). This avoids the interpolation.
> Such
> things are, sadly more expensive.
>
> If I had to buy a new camera, I would try to find out what is supported
> right
> now (some camera manufacturers produce IJ plugins that can drive their
> cameras, but be aware that some of these only support manual acquisition,
> and
> not driven from a plugin or macro, which is really useful).
> Check the MicroManager pages to see what it is supported too.
> If you decide to get a camera supported by an ImageJ plugin but without
> macro
> support, you could still use the IJ_Robot plugin to try to automate it. It
> is
> an ugly way of doing it, but it works.
>
> I hope this is useful.
> Cheers
>
> Gabriel
>

--
Dr W. B. Amos FRS
MRC Laboratory of Molecular Biology
Hills Road, Cambridge CB2 0QH
telephone [phone] (lab)
fax           [phone]
Emails [hidden email]
or [hidden email]
Websites:  (Lab) http://www2.mrc-lmb.cam.ac.uk/SS/Amos_B/
 (Personal) http://homepage.ntlworld.com/w.amos2/

[melkor2.0]

Dear Brad,
I've been reading about Nyquist theorem and so, and I don't get the math, it
seems 3x it's at the maximum limit of the linear region, but how do you
calculate the final total resolution required? you say for 60x - 1.4 we need
0.9 megapixels and to work with 3x linear maginification, that is nine times
this, can you explain or reference the math?
thanks indeed,

ramon

On 12 February 2011 14:22, Brad Amos <[hidden email]> wrote:

> Dear Gabriel,
>                     You will not find resolution values in the scant
> literature supplied by microscope manufacturers, but you can calculate how
> many pixels you will need from the N.A. and magnification written on the
> objective.  Assuming Nyquist sampling, the pixel separation (assuming no
> extra magnification) is  21, 14 and 16 microns for three commonly-used
> objectives (60x N.A. 1.4, 20x N.A. 0.7 and 10x N.A. 0.3). So, if we wish to
> capture a 20 x 20 mm square area of the intermediate image (a reasonably
> large fraction of what we see in an eyepiece)  we need 0.9, 2 and 1.5
> megapixels minimum (i.e. to record full detail without any empty
> magnification. In practice, microscopists usually choose to work with about
> 3x the linear magnification at the Nyquist minimum,  so this means that the
> preferred number of pixels would be nine times this, unless the camera was
> recording a reduced area of the intermediate image. This means that the
> high
> pixel numbers of the modern  DSLRs are not overkill.  This is why standard
> PAL video resolution (about 0.4 megapixels maximum) is so hopeless for
> microscopy, forcing users to capture only a tiny fraction of the eyepiece
> field.
> Brad Amos
>
> On 12 February 2011 12:20, Gabriel Landini <[hidden email]> wrote:
>
> > On Saturday 12 Feb 2011, bradscopegems wrote:
> > > If your requirement is for still images, even a medium-price digital
> SLR
> > > camera such as the Nikon D90 will give superb results at higher
> > resolution
> > > (14 Megapixels).
> >
> > The resolution is mainly driven by the microscope optics. With a 14 MP
> > camera
> > one gets what is called "empty magnification".
> > I would suggest to find out what are the resolution values of the
> > microscope
> > objectives are (from the manufacturer manuals) and start thinking from
> > there.
> > Depending on the field width of the camera and the optics, I think
> anything
> > >
> > 4 MP or thereabouts will not capture much more detail than the resolution
> > of
> > the optical system will be able to provide in a relatively good
> brightfield
> > microscope.
> >
> > Of course one can subsample large images with empty magnification and
> > reduce
> > file size while keeping the useful amount of detail, but there are other
> > disadvantages. With a SLR one will struggle to make illumination
> > modifications
> > on the fly to correct for brightfield background illumination using the
> > transmittance method. You can do it, but it will be more time consuming
> as
> > one
> > cannot check histogram saturation on the fly, etc. then transfer the
> images
> > to
> > the computer, reload them, etc. Whether this is crucial it depends on how
> > many
> > images one is expecting to acquire.
> >
> > Also remember that colour cameras using Bayer masks will interpolate
> > colours
> > and that reduces further the resolution of the image data. A good
> > alternative
> > is to use greyscale cameras with a R-G-B filter wheel or tunable filter
> so
> > each pixel is exposed 3 times (R G & B). This avoids the interpolation.
> > Such
> > things are, sadly more expensive.
> >
> > If I had to buy a new camera, I would try to find out what is supported
> > right
> > now (some camera manufacturers produce IJ plugins that can drive their
> > cameras, but be aware that some of these only support manual acquisition,
> > and
> > not driven from a plugin or macro, which is really useful).
> > Check the MicroManager pages to see what it is supported too.
> > If you decide to get a camera supported by an ImageJ plugin but without
> > macro
> > support, you could still use the IJ_Robot plugin to try to automate it.
> It
> > is
> > an ugly way of doing it, but it works.
> >
> > I hope this is useful.
> > Cheers
> >
> > Gabriel
> >
>
>
>
> --
> Dr W. B. Amos FRS
> MRC Laboratory of Molecular Biology
> Hills Road, Cambridge CB2 0QH
> telephone [phone] (lab)
> fax           [phone]
> Emails [hidden email]
> or [hidden email]
> Websites:  (Lab) http://www2.mrc-lmb.cam.ac.uk/SS/Amos_B/
>  (Personal) http://homepage.ntlworld.com/w.amos2/
>

[Peter van Loon]

Thank you for your reply.


I am really thinking now for a DSLR option. Canon has the Eos utility and Nikon the Camera Control software to capture the image controlled by a PC but I don't have experience on using it. So which one will be best? Of course this software can't be controlled by ImageJ. But it still is an option.

Normally people are using a professional camera normally with C-mount (that's why I mentioned C mount as a spec). But this is not necessary for me. I wont use it for microscopy but by image analysis on plants.
Sorry, I was not clear on that point.

DLSR camera's have color interpretation (Bayer) but most professional camera's for image analysis use it also (exept 3-CCD or camera;s with color-wheel).
So what will be the biggest difference? Noise?

So if I have to choose because of budget reasons between a low resolution professional camera or a high resolution DSLR camera, maybe the last is the best?

Does anyone has experiences with using DSLR camera with image analysis?

Thanks in advance!

Peter van Loon

[Gabriel Landini]

On Saturday 12 Feb 2011, you wrote:
>                      You will not find resolution values in the scant
> literature supplied by microscope manufacturers,

Hi Brad,
The Olympus objectives we use came with some blurb where it was stated.
However, I just checked and strangely they do not provide this information in
their co.uk website. The objectives have their specifications listed but not
resolving power.

> So, if we wish to  capture a 20 x 20 mm square area of the intermediate
> image (a reasonably large fraction of what we see in an eyepiece)  we need
> 0.9, 2 and 1.5 megapixels minimum (i.e. to record full detail without any
> empty magnification.
> In practice, microscopists usually choose to work with about
> 3x the linear magnification at the Nyquist minimum,  so this means that the
> preferred number of pixels would be nine times this, unless the camera was
> recording a reduced area of the intermediate image. This means that the
> high pixel numbers of the modern  DSLRs are not overkill.

My point, (maybe I did not articulate it well) is that the data being stored
in such large number of pixels would not be adding anything in terms of image
detail and yet it will require larger storage and more processing.
If this is not taken into consideration one risks processing and reporting
morphological detail which could not be resolved in the first place. This is
of course obvious to experienced microscopists, but not perhaps to those who
did not think of this in the first place.

Let's use the example of fractal objects imaged with such level of empty
magnification. Applying the yardstick method, their perimeters measured with
small yardstick sizes will appear smoother than they really are because the
detail of sizes close to the image pixels cannot be resolved.

I must confess that I wasn't aware of the 3x preference by microscopists. Is
there a reason for this number?
Regards,

Gabriel

[bradscopegems-2]

Dear Gabriel,
                     The 3x is anecdotal and subjective, but I can account
for it by reference to my previous message about the intensity profile of
the image of two point objects separated by the Rayleigh distance. This
curve is shown in practically every textbook dealing with microscope optics.
At the minimum Nyquist sampling frequency, the intensity  profile across the
resolved region will consist of just three points: two high  points
corresponding to the peaks and a low point in between. Only at about 3x this
frequency will one begin to see something of the shape of the peaks and of
the valley in between. And, of course, a high signal to noise ratio,
requiring many hundreds of photons coming from each resolved point, is
necessary to see the dip at all.
Brad

On 12 February 2011 17:06, Gabriel Landini <[hidden email]> wrote:

> On Saturday 12 Feb 2011, you wrote:
> >                      You will not find resolution values in the scant
> > literature supplied by microscope manufacturers,
>
> Hi Brad,
> The Olympus objectives we use came with some blurb where it was stated.
> However, I just checked and strangely they do not provide this information
> in
> their co.uk website. The objectives have their specifications listed but
> not
> resolving power.
>
> > So, if we wish to  capture a 20 x 20 mm square area of the intermediate
> > image (a reasonably large fraction of what we see in an eyepiece)  we
> need
> > 0.9, 2 and 1.5 megapixels minimum (i.e. to record full detail without any
> > empty magnification.
> > In practice, microscopists usually choose to work with about
> > 3x the linear magnification at the Nyquist minimum,  so this means that
> the
> > preferred number of pixels would be nine times this, unless the camera
> was
> > recording a reduced area of the intermediate image. This means that the
> > high pixel numbers of the modern  DSLRs are not overkill.
>
> My point, (maybe I did not articulate it well) is that the data being
> stored
> in such large number of pixels would not be adding anything in terms of
> image
> detail and yet it will require larger storage and more processing.
> If this is not taken into consideration one risks processing and reporting
> morphological detail which could not be resolved in the first place. This
> is
> of course obvious to experienced microscopists, but not perhaps to those
> who
> did not think of this in the first place.
>
> Let's use the example of fractal objects imaged with such level of empty
> magnification. Applying the yardstick method, their perimeters measured
> with
> small yardstick sizes will appear smoother than they really are because the
> detail of sizes close to the image pixels cannot be resolved.
>
> I must confess that I wasn't aware of the 3x preference by microscopists.
> Is
> there a reason for this number?
> Regards,
>
> Gabriel
>

--
Dr W. B. Amos FRS
MRC Laboratory of Molecular Biology
Hills Road, Cambridge CB2 0QH
telephone [phone] (lab)
fax           [phone]
Emails [hidden email]
or [hidden email]
Websites:  (Lab) http://www2.mrc-lmb.cam.ac.uk/SS/Amos_B/
 (Personal) http://homepage.ntlworld.com/w.amos2/

[Gabriel Landini]

On Saturday 12 Feb 2011, <[hidden email]> wrote:
> Normally people are using a professional camera normally with C-mount
> (that's why I mentioned C mount as a spec).

C mount a common thread in cameras and microscope adaptors. To attach a Nikon
SLR camera to a microscope you need an adaptor called F mount.

> DLSR camera's have color interpretation (Bayer) but most professional
> camera's for image analysis use it also

Some do, some don't

> So what will be the biggest difference? Noise?

I already mentioned this in a previous email. The Bayer mask over the sensor
has green red and blue pixels. To assign the 3 colours to a final "image
pixel", the Bayer mask is interpolated. So in reality, the resolution that the
manufacturer reports is not accurately the one in the acquired image. Have a
look at the wikipedia article on Bayer mask.
http://en.wikipedia.org/wiki/Bayer_mask

Cameras with the wheel can capture R , G and B for each pixel. The colour is
more accurate as it does not need to be the interpolated and so the resolution
is as specified byt number of pixels in the sensor. The problem is that as the
images are taken in sequence the objects should not move between shots. The
tunable filter cameras are quicker than the filter wheel ones (or at least the
one I have is quicker than the wheel ones I saw).

> So if I have to choose because of budget reasons between a low resolution
> professional camera or a high resolution DSLR camera, maybe the last is the
> best?

Which low resolution professional and which DSLR? The question does not make
much sense yet.

> Does anyone has experiences with using DSLR camera with image analysis?

Since you don't say what you are trying to achieve, here are some generic
suggestions.
Do not use jpegs. Use uncompressed (non-lossy) formats for your images (TIFF
or RAW (there is a plugin DC-RAW somewhere to read these).
Use a standardised illumination source so the images are comparable.
Use a fixed focal length in your shots so the magnification is also
comparable.
Add a colour calibration tablet to calibrate the colours and the
magnification, see:
http://imagejdocu.tudor.lu/doku.php?id=plugin:color:chart_white_balance:start
Use the camera in Manual mode so you are in control of all the settings.
Use a tripod.

Cheers

Gabriel

[Robert Dougherty]

Brad,
3x is my experience from developing acoustical imaging systems. In this case the pixels are computed, so it is important to get the spacing right to make the computation efficient.  I've been experimenting with the factor for years, and 3 seems to be the optimum for simple beamforming methods that are limited in resolution by the Rayleigh expression. Thanks for the Nyquist interpretation.
Bob

Robert P. Dougherty
President
OptiNav, Inc.
1414 127th Pl NE #106
Bellevue, WA 98005
[phone]
FAX [phone]
[hidden email]
www.optinav. com

On Feb 12, 2011, at 12:37 PM, Brad Amos <[hidden email]> wrote:

> Dear Gabriel,
>                     The 3x is anecdotal and subjective, but I can account
> for it by reference to my previous message about the intensity profile of
> the image of two point objects separated by the Rayleigh distance. This
> curve is shown in practically every textbook dealing with microscope optics.
> At the minimum Nyquist sampling frequency, the intensity  profile across the
> resolved region will consist of just three points: two high  points
> corresponding to the peaks and a low point in between. Only at about 3x this
> frequency will one begin to see something of the shape of the peaks and of
> the valley in between. And, of course, a high signal to noise ratio,
> requiring many hundreds of photons coming from each resolved point, is
> necessary to see the dip at all.
> Brad
>
> On 12 February 2011 17:06, Gabriel Landini <[hidden email]> wrote:
>
>> On Saturday 12 Feb 2011, you wrote:
>>>                     You will not find resolution values in the scant
>>> literature supplied by microscope manufacturers,
>>
>> Hi Brad,
>> The Olympus objectives we use came with some blurb where it was stated.
>> However, I just checked and strangely they do not provide this information
>> in
>> their co.uk website. The objectives have their specifications listed but
>> not
>> resolving power.
>>
>>>
>>
>>
>>
>>
>>

[David Gene Morgan]

Hi,

        I'm responding to the question of whether there is a reason for the 3x
sampling reported to be commonly used by microscopists.  Since no-one
else has brought the following up, I decided it might be useful in this
discussion (not in terms of buying a camera, but useful in the sense of
being informative to the discussion).  And I'll apologize in advance if
this seems too obvious and/or pedantic for anyone/everyone.

        To start, forget about the optics, the sensor and everything else about
producing a digital image, and only think about the image as a digital
signal.  For this purpose, you can even assume that everything leading
up to the image itself is "perfect" and that the only loss of
resolution/signal when going from the original object to the image is
the fact that the final image is digital (not sampled using infinitely
small steps).

        When one does this, the "3x" sampling referred to in previous postings
becomes relatively easy to explain, and not so anecdotal and subjective.
  People have mentioned the Nyquist sampling frequency but no-one has
gone into any detail about it.  This concept comes from information
theory and the microscopy community that I am familiar with (electron
microscopy) lumps a lot of not terribly rigorous ideas about it into
phrases such as the Nyquist limit, Shannon sampling (named after Claude
Shannon, one of the fathers of information theory), etc.

        There is a lot of information about these ideas in various places on
Wikipedia, and to quote from there, the Nyquist-Shannon sampling theorem
says:

        If a function x(t) contains no frequencies higher than B hertz, it is
completely determined by giving its ordinates at a series of points
spaced 1/(2B) seconds apart.

        The inverse of this is also true:  if the sampling is 1/(2B), the
maximum possible frequency contained in a (sampled) function y(t) is B
hertz.

        The word "possible" is important in that last sentence:  the original
function x(t) may or may not have frequencies higher than B hertz, but
the function y(t), sampled at 1/(2B), CANNOT have any frequencies higher
than B (that's the point of the theorem).  Any frequencies that were
originally present in x(t) at frequencies lower than B will be preserved
in the sampled signal, and if there were not any frequencies lower than
B (a totally aperiodic signal, for example), the sampled function will
also not have any lower frequencies...

        For a digital image, the "function" is the image itself (a 2d function,
but that doesn't matter), the sampling is the pixel size (how much of
the original object is represented by a pixel in the digital image, a
spatial instead of a temporal sampling) and so the maximum resolution
that can be contained in a digital image with pixel size N is 1/(2N).
This has nothing to do with the number of pixels in an image, and
everything to do with the pixel size itself (and note that this is not
the pixel size of the sensor, but the apparent pixel size in the image).

        One way to have this make sense is to think about resolution as meaning
"the ability to distinguish that two features are indeed two features
and not one."  This is related to the Rayleigh criterion, and brings
into the discussion things like numerical aperture when dealing with the
resolution of lenses and cameras.  But since we are assuming that all
this is perfect at the moment, all we care about is whether we can tell
two objects from one in the final digital image.

        If the features are close enough, a digital image will show them as a
single large feature (i.e., it can't resolve the large feature into two
separate, smaller things).  In order to tell that there are really two
features beside one another in an image, one needs at a minimum a single
pixel between the two features that clearly belongs to neither.  If the
pixel size is small enough, there will be a pixel between the two that
distinguishes one from the other, and it is this "second pixel" as one
travels from digital point to digital point that says the resolution is
sufficient to tell the two features are not a single large "thing."

        At one extreme, if the "features" are only a pixel wide (or better
said, if the signal from a feature can be contained in a single pixel),
this all means that one needs 2 pixels (in all directions, but think
about a 1d case in this context) to determine that one is seeing a
single, isolated feature and not two adjacent ones:  the first pixel has
the signal, the second pixel says that signal isn't there and the third
pixel could either say a signal is still not there or that there is a
new signal present (the second feature). To put this back into the
framework of Nyquist-Shannon sampling, the point-like features we want
to see have size x (equal to a pixel in the digital image) and the
resolution that defines the ability to tell that there are two such
features side by side is 1/(2x).  This is all a TERRIBLY non-rigorous
description of the Nyquist-Shannon sampling theorem but it makes the
point reasonably well and fits with some of what has been said before.

        This all might seem to indicate that the "3x" value noted in previous
postings should be "2x" instead and indeed, the maximum possible
resolution in a digital image is 1/(2x) (the Nyquist-Shannon sampling
theorem again).  However, if a digital image is manipulated in just
about any way that affects its sampling (pixellation), the image loses
information.  For example, transformations such as rotation by amounts
that aren't multiples of 90 degrees, or shifts in x and/or y by
fractions of a pixel, require an interpolation of the image, and that
interpolation leads to information loss.  Another way to describe this
is to say that the original maximum resolution of 1/(2x) is degraded
after such image transformations even though the sampling is still x.

        This in turn means that in the case where the resolution of a digital
image is at the limit of what the user needs in order to see a feature,
after relatively common image transformations, the resolution will no
longer be sufficient and the feature(s) will vanish.

        A reasonable solution to this is to over-sample so that the desired
resolution is well within the 1/(2x) limit imposed by sampling using
steps of x.  To give a concrete example, if the pixels in a digital
image represent 500 nm, the maximum resolution is 1/(2*500) and one can
just distinguish 1000 nm features.  If the goal is to see such features,
it would be better NOT to live on the edge of detectability and to
sample at ~330 nm/ pixel (so that a resolution limit of 1/(3x)  allows
the 1000 nm features to be seen), or even 250 nm (where 1/(4x) gives the
desired resolution to see a 1000 nm feature).  In this case, we often
say that we want 1000 nm resolution, and in order to obtain that, we
need to chose our sampling distance x so that 3x (or even 4x) equals
1000 nm (instead of the absolute minimum of 2x).

        One could obviously over-sample by a factor of (say) 5 and not even
think about sampling issues having an effect on feature resolution.
However, the 5x over-sampled image has 25x more pixels, and takes 25x
more computer memory, storage space, etc.  This massive over-sampling
can be the "empty resolution" that people sometimes mention (though
other types of empty resolution are also possible).   Over-sampling at
1.5x only increases things by a small factor (2.25x) and still results
in virtually no loss of feature resolution (and no empty resolution).

        The above discussion is certainly the root of the of "3x sampling"
commonly mentioned in the cryo-electron microscopy field, and I suspect
it plays a major part in the sampling schemes of most other digital
imaging processes.  LOTS of other things affect the resolution in a
digital image, and these can be as important as what is discussed here.
  But since it is theoretically impossible to get resolution that is
better than 1/(2x) from an image sampled in steps of x, one is most
often better served by sampling a bit more finely:  If one wants a
resolution of 1/Q, sample at Q/3 instead of Q/2!

Brad Amos wrote:

> Dear Gabriel,
>                      The 3x is anecdotal and subjective, but I can account
> for it by reference to my previous message about the intensity profile of
> the image of two point objects separated by the Rayleigh distance. This
> curve is shown in practically every textbook dealing with microscope optics.
> At the minimum Nyquist sampling frequency, the intensity  profile across the
> resolved region will consist of just three points: two high  points
> corresponding to the peaks and a low point in between. Only at about 3x this
> frequency will one begin to see something of the shape of the peaks and of
> the valley in between. And, of course, a high signal to noise ratio,
> requiring many hundreds of photons coming from each resolved point, is
> necessary to see the dip at all.
> Brad
>
> On 12 February 2011 17:06, Gabriel Landini <[hidden email]> wrote:
>
>> On Saturday 12 Feb 2011, you wrote:
>>>                      You will not find resolution values in the scant
>>> literature supplied by microscope manufacturers,
>> Hi Brad,
>> The Olympus objectives we use came with some blurb where it was stated.
>> However, I just checked and strangely they do not provide this information
>> in
>> their co.uk website. The objectives have their specifications listed but
>> not
>> resolving power.
>>
>>> So, if we wish to  capture a 20 x 20 mm square area of the intermediate
>>> image (a reasonably large fraction of what we see in an eyepiece)  we
>> need
>>> 0.9, 2 and 1.5 megapixels minimum (i.e. to record full detail without any
>>> empty magnification.
>>> In practice, microscopists usually choose to work with about
>>> 3x the linear magnification at the Nyquist minimum,  so this means that
>> the
>>> preferred number of pixels would be nine times this, unless the camera
>> was
>>> recording a reduced area of the intermediate image. This means that the
>>> high pixel numbers of the modern  DSLRs are not overkill.
>> My point, (maybe I did not articulate it well) is that the data being
>> stored
>> in such large number of pixels would not be adding anything in terms of
>> image
>> detail and yet it will require larger storage and more processing.
>> If this is not taken into consideration one risks processing and reporting
>> morphological detail which could not be resolved in the first place. This
>> is
>> of course obvious to experienced microscopists, but not perhaps to those
>> who
>> did not think of this in the first place.
>>
>> Let's use the example of fractal objects imaged with such level of empty
>> magnification. Applying the yardstick method, their perimeters measured
>> with
>> small yardstick sizes will appear smoother than they really are because the
>> detail of sizes close to the image pixels cannot be resolved.
>>
>> I must confess that I wasn't aware of the 3x preference by microscopists.
>> Is
>> there a reason for this number?
>> Regards,
>>
>> Gabriel
>>
>
>
>

--
                  David Gene Morgan
                   cryoEM Facility
                   320C Simon Hall
            Indiana University Bloomington
                812 856 1457 (office)
                812 856 3221 (EM lab)
             http://bio.indiana.edu/~cryo

[Chu, Calvin-2]

Our lab bought a new Tucsen cooled CCD camera for like $700 USD or something like that through ebay.

Be weary of purchasing lab equipment through ebay though b/c it is most likely be stolen.

We still haven't tested it out yet though.  The cooling element is a fan though, I prefer thermoelectrics especially for extremely high resolution images.
________________________________________
From: ImageJ Interest Group [[hidden email]] On Behalf Of David Gene Morgan [[hidden email]]
Sent: Monday, February 14, 2011 5:41 PM
To: [hidden email]
Subject: Re: affordable camera suggestions

Hi,

        I'm responding to the question of whether there is a reason for the 3x
sampling reported to be commonly used by microscopists.  Since no-one
else has brought the following up, I decided it might be useful in this
discussion (not in terms of buying a camera, but useful in the sense of
being informative to the discussion).  And I'll apologize in advance if
this seems too obvious and/or pedantic for anyone/everyone.

        To start, forget about the optics, the sensor and everything else about
producing a digital image, and only think about the image as a digital
signal.  For this purpose, you can even assume that everything leading
up to the image itself is "perfect" and that the only loss of
resolution/signal when going from the original object to the image is
the fact that the final image is digital (not sampled using infinitely
small steps).

        When one does this, the "3x" sampling referred to in previous postings
becomes relatively easy to explain, and not so anecdotal and subjective.
  People have mentioned the Nyquist sampling frequency but no-one has
gone into any detail about it.  This concept comes from information
theory and the microscopy community that I am familiar with (electron
microscopy) lumps a lot of not terribly rigorous ideas about it into
phrases such as the Nyquist limit, Shannon sampling (named after Claude
Shannon, one of the fathers of information theory), etc.

        There is a lot of information about these ideas in various places on
Wikipedia, and to quote from there, the Nyquist-Shannon sampling theorem
says:

        If a function x(t) contains no frequencies higher than B hertz, it is
completely determined by giving its ordinates at a series of points
spaced 1/(2B) seconds apart.

        The inverse of this is also true:  if the sampling is 1/(2B), the
maximum possible frequency contained in a (sampled) function y(t) is B
hertz.

        The word "possible" is important in that last sentence:  the original
function x(t) may or may not have frequencies higher than B hertz, but
the function y(t), sampled at 1/(2B), CANNOT have any frequencies higher
than B (that's the point of the theorem).  Any frequencies that were
originally present in x(t) at frequencies lower than B will be preserved
in the sampled signal, and if there were not any frequencies lower than
B (a totally aperiodic signal, for example), the sampled function will
also not have any lower frequencies...

        For a digital image, the "function" is the image itself (a 2d function,
but that doesn't matter), the sampling is the pixel size (how much of
the original object is represented by a pixel in the digital image, a
spatial instead of a temporal sampling) and so the maximum resolution
that can be contained in a digital image with pixel size N is 1/(2N).
This has nothing to do with the number of pixels in an image, and
everything to do with the pixel size itself (and note that this is not
the pixel size of the sensor, but the apparent pixel size in the image).

        One way to have this make sense is to think about resolution as meaning
"the ability to distinguish that two features are indeed two features
and not one."  This is related to the Rayleigh criterion, and brings
into the discussion things like numerical aperture when dealing with the
resolution of lenses and cameras.  But since we are assuming that all
this is perfect at the moment, all we care about is whether we can tell
two objects from one in the final digital image.

        If the features are close enough, a digital image will show them as a
single large feature (i.e., it can't resolve the large feature into two
separate, smaller things).  In order to tell that there are really two
features beside one another in an image, one needs at a minimum a single
pixel between the two features that clearly belongs to neither.  If the
pixel size is small enough, there will be a pixel between the two that
distinguishes one from the other, and it is this "second pixel" as one
travels from digital point to digital point that says the resolution is
sufficient to tell the two features are not a single large "thing."

        At one extreme, if the "features" are only a pixel wide (or better
said, if the signal from a feature can be contained in a single pixel),
this all means that one needs 2 pixels (in all directions, but think
about a 1d case in this context) to determine that one is seeing a
single, isolated feature and not two adjacent ones:  the first pixel has
the signal, the second pixel says that signal isn't there and the third
pixel could either say a signal is still not there or that there is a
new signal present (the second feature). To put this back into the
framework of Nyquist-Shannon sampling, the point-like features we want
to see have size x (equal to a pixel in the digital image) and the
resolution that defines the ability to tell that there are two such
features side by side is 1/(2x).  This is all a TERRIBLY non-rigorous
description of the Nyquist-Shannon sampling theorem but it makes the
point reasonably well and fits with some of what has been said before.

        This all might seem to indicate that the "3x" value noted in previous
postings should be "2x" instead and indeed, the maximum possible
resolution in a digital image is 1/(2x) (the Nyquist-Shannon sampling
theorem again).  However, if a digital image is manipulated in just
about any way that affects its sampling (pixellation), the image loses
information.  For example, transformations such as rotation by amounts
that aren't multiples of 90 degrees, or shifts in x and/or y by
fractions of a pixel, require an interpolation of the image, and that
interpolation leads to information loss.  Another way to describe this
is to say that the original maximum resolution of 1/(2x) is degraded
after such image transformations even though the sampling is still x.

        This in turn means that in the case where the resolution of a digital
image is at the limit of what the user needs in order to see a feature,
after relatively common image transformations, the resolution will no
longer be sufficient and the feature(s) will vanish.

        A reasonable solution to this is to over-sample so that the desired
resolution is well within the 1/(2x) limit imposed by sampling using
steps of x.  To give a concrete example, if the pixels in a digital
image represent 500 nm, the maximum resolution is 1/(2*500) and one can
just distinguish 1000 nm features.  If the goal is to see such features,
it would be better NOT to live on the edge of detectability and to
sample at ~330 nm/ pixel (so that a resolution limit of 1/(3x)  allows
the 1000 nm features to be seen), or even 250 nm (where 1/(4x) gives the
desired resolution to see a 1000 nm feature).  In this case, we often
say that we want 1000 nm resolution, and in order to obtain that, we
need to chose our sampling distance x so that 3x (or even 4x) equals
1000 nm (instead of the absolute minimum of 2x).

        One could obviously over-sample by a factor of (say) 5 and not even
think about sampling issues having an effect on feature resolution.
However, the 5x over-sampled image has 25x more pixels, and takes 25x
more computer memory, storage space, etc.  This massive over-sampling
can be the "empty resolution" that people sometimes mention (though
other types of empty resolution are also possible).   Over-sampling at
1.5x only increases things by a small factor (2.25x) and still results
in virtually no loss of feature resolution (and no empty resolution).

        The above discussion is certainly the root of the of "3x sampling"
commonly mentioned in the cryo-electron microscopy field, and I suspect
it plays a major part in the sampling schemes of most other digital
imaging processes.  LOTS of other things affect the resolution in a
digital image, and these can be as important as what is discussed here.
  But since it is theoretically impossible to get resolution that is
better than 1/(2x) from an image sampled in steps of x, one is most
often better served by sampling a bit more finely:  If one wants a
resolution of 1/Q, sample at Q/3 instead of Q/2!

Brad Amos wrote:

> Dear Gabriel,
>                      The 3x is anecdotal and subjective, but I can account
> for it by reference to my previous message about the intensity profile of
> the image of two point objects separated by the Rayleigh distance. This
> curve is shown in practically every textbook dealing with microscope optics.
> At the minimum Nyquist sampling frequency, the intensity  profile across the
> resolved region will consist of just three points: two high  points
> corresponding to the peaks and a low point in between. Only at about 3x this
> frequency will one begin to see something of the shape of the peaks and of
> the valley in between. And, of course, a high signal to noise ratio,
> requiring many hundreds of photons coming from each resolved point, is
> necessary to see the dip at all.
> Brad
>
> On 12 February 2011 17:06, Gabriel Landini <[hidden email]> wrote:
>
>> On Saturday 12 Feb 2011, you wrote:
>>>                      You will not find resolution values in the scant
>>> literature supplied by microscope manufacturers,
>> Hi Brad,
>> The Olympus objectives we use came with some blurb where it was stated.
>> However, I just checked and strangely they do not provide this information
>> in
>> their co.uk website. The objectives have their specifications listed but
>> not
>> resolving power.
>>
>>> So, if we wish to  capture a 20 x 20 mm square area of the intermediate
>>> image (a reasonably large fraction of what we see in an eyepiece)  we
>> need
>>> 0.9, 2 and 1.5 megapixels minimum (i.e. to record full detail without any
>>> empty magnification.
>>> In practice, microscopists usually choose to work with about
>>> 3x the linear magnification at the Nyquist minimum,  so this means that
>> the
>>> preferred number of pixels would be nine times this, unless the camera
>> was
>>> recording a reduced area of the intermediate image. This means that the
>>> high pixel numbers of the modern  DSLRs are not overkill.
>> My point, (maybe I did not articulate it well) is that the data being
>> stored
>> in such large number of pixels would not be adding anything in terms of
>> image
>> detail and yet it will require larger storage and more processing.
>> If this is not taken into consideration one risks processing and reporting
>> morphological detail which could not be resolved in the first place. This
>> is
>> of course obvious to experienced microscopists, but not perhaps to those
>> who
>> did not think of this in the first place.
>>
>> Let's use the example of fractal objects imaged with such level of empty
>> magnification. Applying the yardstick method, their perimeters measured
>> with
>> small yardstick sizes will appear smoother than they really are because the
>> detail of sizes close to the image pixels cannot be resolved.
>>
>> I must confess that I wasn't aware of the 3x preference by microscopists.
>> Is
>> there a reason for this number?
>> Regards,
>>
>> Gabriel
>>
>
>
>

--
                  David Gene Morgan
                   cryoEM Facility
                   320C Simon Hall
            Indiana University Bloomington
                812 856 1457 (office)
                812 856 3221 (EM lab)
             http://bio.indiana.edu/~cryo

[Janne Hyötylä]

Hi all,

On Mon, 14 Feb 2011 23:41:38 +0100, David Gene Morgan  
<[hidden email]> wrote:

        One could obviously over-sample by a factor of (say) 5 and not even
> think about sampling issues having an effect on feature resolution.  
> However, the 5x over-sampled image has 25x more pixels, and takes 25x  
> more computer memory, storage space, etc.  This massive over-sampling  
> can be the "empty resolution" that people sometimes mention (though  
> other types of empty resolution are also possible).   Over-sampling at  
> 1.5x only increases things by a small factor (2.25x) and still results  
> in virtually no loss of feature resolution (and no empty resolution).


With today's computers the memory and storage issues are probably not so  
important anymore.
I think the most critical issue is that with more pixels (ie. smaller  
pixel size) the amount of photons per pixel decreases, and this decreases  
the signal-to-noise ratio.



> The above discussion is certainly the root of the of "3x sampling"  
> commonly mentioned in the cryo-electron microscopy field, and I suspect  
> it plays a major part in the sampling schemes of most other digital  
> imaging processes.  LOTS of other things affect the resolution in a  
> digital image, and these can be as important as what is discussed here.  
>   But since it is theoretically impossible to get resolution that is  
> better than 1/(2x) from an image sampled in steps of x, one is most  
> often better served by sampling a bit more finely:  If one wants a  
> resolution of 1/Q, sample at Q/3 instead of Q/2!

You can also see a geometrical derivation that leads to a minimal sampling  
of 2.8x optical resolution in Pawley's Handbook of biological confocal  
microscopy:
<http://books.google.com/books?id=IKcPnaNPrhoC&lpg=PA64&ots=r8SqLx2avr&dq=nyquist%202d%20sampling&pg=PA65#v=onepage&q&f=false>

I have also seen 2.3x mentioned that was apparently also somehow derived  
 from Nyquist for 2D signals, but without the actual derivation.

So while 3x might be "anecdotal", it is not far away from mathematically  
derived values.

Best,
Janne

[Gabriel Landini]

On Monday 14 Feb 2011 22:41:38 David Gene Morgan wrote:
>       I'm responding to the question of whether there is a reason for the 3x
> sampling reported to be commonly used by microscopists.  

I think both Brad's and your explanations were very useful, thank you.

There seem to be a variety of possible combinations of extra optics in the
extension tubes that one could add: x0.25, x0.5, x0.75, x1, x1.2, 1.5, x2.5,
x3.3, x4, x5.
Presumably these also have an impact on the resolution of the
image projected on the sensor. Is that just a multiplicative effect only, or
it there further deterioration of the image resolution due to these?

On a related note, I found the information of the objectives resolution (in
the Olympus BX50 manual, page 24). Given that this information is, curiously,
not in the Olympus website, if anybody needs it, just send me a private mail.

Regards,

Gabriel

[Philippe GENDRE]

Dear Peter,

My experience is the following. For a full compatibility with Image J you
have two solutions :
- cameras with device adapters through Micro-Manager see
http://valelab.ucsf.edu/~nico/MMwiki/index.php/Device_Support; Micro-Manager
can be seen as an ImageJ plugin driving hardwares,
- the FireCam plugin see
http://www.phase-hl.com/cgi-bin/querypage.cgi?FireCamIJ_uk

Hoping this helps you.

Philippe*
*
2011/2/11 Peter van Loon <[hidden email]>

> Dear All,
>
> I am looking for a camera compatible with ImageJ.
> It should have the next specs: at least 2-3 Mega Pixel, colour and C-mount
> Surfing the internet I see only expensive solutions. Does anyone know some
> affordable solutions (2000-3000 euro max),
>
> Thanks in advance!,
>
> Peter van Loon
>
> The information contained in this e-mail (and attachments if any) is
> exclusively intended for the recipient(s) named above and may be
> confidential, proprietary, and/or legally privileged. Unintentional
> disclosure of this e-mail does not constitute a waiver of any right or
> privilege. Please notify us if you are not the intended recipient and do not
> use, print, copy, forward, or disclose (any part of) this e-mail, but delete
> it (and attachments and copies if any) subsequently. Thank you.
>
> You can find disclaimer translations in different languages by visiting our
> website http://www.rijkzwaan.com/edisclaimer
>
> The registration number of each Rijk Zwaan company can be found by using
> the following link:
> http://www.rijkzwaan.com/registrationnumbers
>

[Martin Kielhorn]

2011/2/15 Gabriel Landini <[hidden email]>:

> On Monday 14 Feb 2011 22:41:38 David Gene Morgan wrote:
>>       I'm responding to the question of whether there is a reason for the 3x
>> sampling reported to be commonly used by microscopists.
>
> I think both Brad's and your explanations were very useful, thank you.
>
> There seem to be a variety of possible combinations of extra optics in the
> extension tubes that one could add: x0.25, x0.5, x0.75, x1, x1.2, 1.5, x2.5,
> x3.3, x4, x5.
> Presumably these also have an impact on the resolution of the
> image projected on the sensor. Is that just a multiplicative effect only, or
> it there further deterioration of the image resolution due to these?

In a microscope the tube length (distance between principal planes of
objective and tubelens)
should be the sum of the focal length of the tubelens (f_Tl=164.5 mm
for most Zeiss systems)
and the focal length of the objective (f_obj=164.5/MAG=2.61 mm for a
magnification of 63).

If this condition is fulfilled the system is called 'telecentric' and
has the property that features
in different z-slices don't change their size, when you focus through
the sample.

In our Zeiss Axiovert 200M the tube lenses (Zeiss calls them Optovar)
1.0x, 1.8x and 2.5x are
all in one turret at the same distance from the objective (presumably
164.5 mm). These lenses
look like singlets with small curvature, so I think the principal
planes are close to the center of the lens.
This, however, means that the system is telecentric only for an Optovar 1.0x.

The Optovar 2.5x should actually be in a distance of 2.5 * 164.5 mm =
41 cm from the objective.
I have never figured out a way to quantify the effect of
non-telecentricity on the images.


Regards, Martin

--
Martin Kielhorn
Randall Division of Cell & Molecular Biophysics
King's College London, New Hunt's House
Guy's Campus, London SE1 1UL, U.K.
tel: [phone],  fax: [phone]

[Martin Kielhorn]

Hi,
1) I would like to bring an argument why the optimium sampling rate is
slightly higher
than the Nyquist rate but not at three times the bandwidth of the signal.

2) I will also indicate how you check if you adjusted your
magnification correctly.

3) I give some more information regarding finite sensor pixel size.

In every system that measures light intensity you have noise. Even
with an ideal noise-free
detector light consists of photons. And a measurement of photons is
always plagued by
shot noise. Your signal to noise ratio increases when you detect more
photons. Brad and Dave
suggest to sample at three times the bandwidth of the signal. But that
means they have to
use a smaller detector size, thus increasing the noise in the signal.
They have not increased
information content but increased noise. That is certainly not optimal.

Other explanations in this thread look at features in real space. E.g.
determine the minimum
between two spots which are just resolved. You should digitally
resample the captured image into
a bigger version by zero-padding the Fourier transform. This
interpolation doesn't add noise and you can see all
features magnified or rotate your image.

These were most of my arguments for 1).

Now I describe 2). (I'm sorry that I'm not proficient enough in ImageJ
to give a step by step instruction)

a) Start with a quadratic image you have captured with the microscope
(I suggest 512x512 or 1024x1024).

b) Make sure the edges all have the same brightness.
The discrete Fourier transform is circular and jumps on the edges will
lead to artifacts. Alternative ways to achieve that:

b1) If you use a fluorescent microscope move a few sub-resolution beads (170nm
diameter yellow green for 63x oil immersion) into the center of the
field of view.
Capture a dark image (without illumination) as well and subtract it
from the bead image.

b2) If you use a bright field microscope capture an bright image A
without sample, an image
of the sample B that only fills the center of the field of view and a
dark image C.
Calculate (A-C)/(B-C). This calculation will be correct illumination
non-uniformity.

b3) Multiply with a window that is one in the center gradually reaches
zero at the
border of the image.

b4) Just try without correction and live with the artifacts.


c) Do a Fourier transform of the image and look at the log of the magnitude.

If you oversampled the band limited signal you should see a disk with a diameter
smaller than the Fourier image around the center.

If you undersampled the signal the disk will either hit the edges and
fold back into the Fourier image (aliasing) or information is everywhere and the
Fourier image is bright everywhere (in this case you could try to
defocus the image
a little bit).

For sampling at the Nyquist limit the disk fits perfectly into the image.


If you happen to use bright field or you have an objective with an adjustable
aperture you can try decreasing the diameter of the aperture. Your image should
turn darker, lines should blur and in the Fourier transform of the
image the disk
should be smaller.



3)
The other posters haven't referred to the fact that the camera pixel is of
a certain size. If you speak about sampling distance, you should refer to the
pixel pitch P -- the distance between the mid points of two sensor
pixels. The pixel
size S is the light sensitive array and related to the fill factor G
of the sensor (or
the sensor and microlenses). In general the pixel size is smaller than
the pixel pitch.

First think of the problem in 1D: Sound cards usually contain a 'sample and
hold circuit' (http://en.wikipedia.org/wiki/Sample_and_hold) that freezes the
voltage u(t) at a certain point t_0 in time so that the ADC can digitize the
value. In a camera this corresponds to a pixel that is infinitesimally small and
therefore wouldn't collect much light.

The question I will now try to answer is: "What happens to the sampled signal
due to a finite sensor size?"

The effect can be understood by looking into Fourier space.
The image is sampled by a 2d grid with periodicity P.
The Fourier transform of a grid is another grid
http://en.wikipedia.org/wiki/Dirac_comb .

One can introduce finite sensor size by convolving the sensors grid
of period P with the rectangle of size S. In the Fourier transform
this results (convolution theorem) in a multiplication by a sinc
function (the Fourier transform of a square).
The sinc function is 1 in the center, i.e. the average pixel intensity
of the whole image is the same, independent on pixel size S.

On all other points of the Fourier plane the sinc function is smaller than 1.
That means that the integrating effect of the finite pixel size reduces the
cameras ability to capture high frequency content. Fortunately the reduction
is moderate even for a fill factor of 100% with pixel pitch equal to
pixel pitch P=S.

If you know the geometry of your sensor you could correct for this
effect but I haven't
seen anyone doing this for imaging.
Maybe it is used when people capture holograms with cameras. That is the problem
where I first thought about the effect of finite pixel size.

Regards, Martin

--
Martin Kielhorn
Randall Division of Cell & Molecular Biophysics
King's College London, New Hunt's House
Guy's Campus, London SE1 1UL, U.K.
tel: [phone],  fax: [phone]