Re: camera calibration

[Jan Hollan <jhollan@amper....muni.cz>]

Dear Dr. Verkaik,

I am sorry I did not made it to write any thorough text in English
by the time of Symposium (in spite of promising to do so). So, there are
but my lecture from the last year

  http://amper.ped.muni.cz/light/lectures/06IAU50t_small.pdf

with some links, the best being perhaps the poster

  http://amper.ped.muni.cz/noc/english/canc_rhythm/g_camer.pdf

and the luminance measurement directory with its README.html at the
bottom:

  http://amper.ped.muni.cz/light/luminance/

Basically, a straightforward non-astronomical calibration may be a
full-size fish-eye image and a luxmeter reading (they should give the same
value, using a proper constant of proportionality between pixel readings
and luminance; however, vignetting and geometry of the image is to be
known ahead). A proxy is a small light source (like a window in a dark
room, I've employed my roofspace at home) imaged by lens, taking two
luxmeter readings, one with the shadow cast onto the luxmeter sensor (than
from the image, the luminance of the source times cos elongation from
the lens axis is integrated).

Historically, I began the calibration by imaging a white paper (assuming
its albedo to be 0.87) and computing its solar or lunar illuminance
astronomically.  Then I got an albedo standard (Labsphere's Spectralon,
0.99) and three luxmeters, so I could measure its illuminance and got its
luminance (illuminance * 0.99 / pi).

Using BaSO4 powder (or a poor water paint from it, or a more adhesive
one using polyvinylalcohol) with about 0.96 albedo is a cheap alternative.
Or even latex-powder mixture, see
http://www.usu.edu/cpl/PDF/Barium_Sulfate.pdf (Reflectance of Barium
Sulfate and White Paint Mixtures). (Some broad band solar values are
   The Solar-AC FAQ: Table of absorptivity and emissivity
           of common materials and coatings
   URL: http://www.solarmirror.com/fom/fom-serve/cache/43.html ).

Perhaps the easiest way is using any luxmeter and any large non-coloured
facade, on overcast days. Luxmeter reading taken toward the facade gives
its luminance * pi. The camera its luminance in arbitrary units. Converting
them to cd/m2 needs just dividing these two values. I've employed fresh
snow for this purpose as well.

Ideally, all possible calibration methods should be employed, and give the
same results. They matched very well for the cameras I've calibrated.

(Of course I should write a thorough paper on my radiometry methods and
raw2lum programme and auxiliary scripts at last... I'm sorry to be so late
with that. I've described it extensively just in Czech, and even this is
by far not ideal, the text missing any images, graphs etc., unlike the
Prague presentation).

with best regards
 Jenik H