NavList:

   

Reply

On 2015-01-07 11:21, Greg Rudzinski wrote:
> A revisit to my first unsuccessful efforts (8 years or so ago) at point and 
shoot camera CN using a Canon Powershot 10MP has me thinking that there may 
be a way to bring this type of affordable camera into the relm of practical 
CN. An artifitial horizon will be used for calibration observations to 
construct a table of minutes of arc per pixel at various pixel counts.

Note that the number of pixels per degree is not constant across the
image. Consider the case of a shot perpendicular to a brick wall. If the
lens is free of distortion, it projects an exact miniature of the wall
on the image plane. That is, bricks are the same size (pixels)
everywhere in the image. But their angular dimensions as seen from the
camera position are quite variable, depending on the angle between the
line of sight and the wall plane. However, the point may be moot if the
photo has a small angle of view and maximum accuracy is not necessary.

The transformation between spherical coordinates in the sky and
rectangular coordinates on a photographic plate is explained in books
such as "Textbook on Spherical Astronomy" by Smart (rev. by Green in
1977). Green also covered the topic in his own 1985 book. The math is
easily programmed in a calculator.

My attempt to attain accuracy on the order of a sextant with a DSLR was
disappointing. Despite a third degree polynomial to compensate for lens
distortion, I fell well short. To be fair, I intentionally made the test
difficult. The zoom lens was set for a moderate wide angle - about the
equivalent of a 35 mm lens on a 35 mm film camera. And I used one set of
distortion coefficients for the full width of the frame. Accuracy would
have been far higher if I'd worked with a small portion of the frame, as
one would normally do if measuring the coordinates of some object with
respect to identified stars.

   

Reply