thermal imagers maximum resolution

On Aug 19, 8:44*pm, Thomas Womack
wrote:
In article ,
Ron Cuaz wrote:

Is this an accurate example of an actual setup.. that is.. where the
circle of confusion or Airy disc is 25 micron for a typical thermal
imager?

I was using figures for focal length and aperture from an optical
camera because I couldn't at the time find the figures for an IR one.

Now I have found them. *The FLIR i5 has a focal length of 6.76
millimetres, a field of view of 17 degrees, and an F-number of 1.5
(that is, an effective aperture of 6.76/1.5 = 4.5 millimetres).

Which gives an Airy-disc size of 0.00676 * 1.22 * 0.00001 / 0.0045 =
18 microns.


Hmm... I think we made a major mistake here. The diameter of the Airy-
disc is 2.44 (wavelength)(f/D) which would make it 36 microns!

This means half that size (for 1/2 sampling formula) or 18 micron is
the smallest pixel size that can be use for thermal imager. Anything
smaller and it's a waste of resolution. What this means is that the
current resolution we have (of 320x240 handheld or 640x480 in bigger
format) is already the optimum one. Note most thermal camera specs
have focal ratio of 1.5 (this is to make the spot small). So to
increase the resolution, the lens have to be enlarge as well as the
sensor chip. If anyone who doesn't agree with this conclusion, please
give the arguments.




If 1024x768 is at 17 um pixels... then 2 megapixel would be
less than 10 um.. and since it can't be smaller than 10um.. then 2
megapixel is the limit for hand held thermal camera, right?

You could go to physically bigger sensors; 1024x768 with 17um pixels
is a sensor about 18mm x 13mm, but you could (at significant expense)
make a 2048x1536 array with 17um pixels which would be a sensor the
size of the one in a DSLR.

There is something I don't quite get. It's the field of view. For
example. The FLIR i5 has field of view of 17 degree x 17 degree at
80x80 resolution while FLIR i7 has a field of view of 25 degree x 25
degree at 120x120 resolution.

That sounds as if the i5 and i7 use pixels of the same size, and
probably the same optics, but the i7 has a physically bigger detector;
the fact that the i5 and i7 have the same 'spatial resolution' of
3.71mrad confirms this.

3.71 milliradians with a focal length of 6.76 millimetres corresponds
to 25-micron pixel size; that is, the i5 has a detector two
millimetres on a side covered in 25-micron pixels, and the i7 has a
detector three millimetres on a side covered in 25-micron pixels.

This means if you have both i5 and i7 side by side aiming at a target
a certain distance away, the view from the i5 screen would be more
magnified given both housing or built are the same?

Assuming the screens show one pixel-on-screen per pixel-on-detector,
the object would look the same size on the i5 and i7 screens, you'd
just see more around it on the i7.

But it's possible that the i5 uses three pixels-on-screen per
pixel-on-detector and the i7 uses two, in which case the view on the
i5 screen would look 50% bigger.

Tom