Internal: Faking fiducial marks for air photo orthorectification

>>Hi BSC,
>>
>>Let me just start off by apologizing for this question:
>>
>>I have a customer who wants to know mathematically how the fiducial
>>marks are used to build an interior orientation in orthorectification.
>>I have explained the role of the fiducial marks and the interior
>>orientation in a qualitative way, but I can't address the mathematics.
>>
>>There is a long and ugly story behind this question. He might be
>>satisfied with an explanation of how complicated and difficult the model
>>is. Basically he wants to know (for the 5th time) why he can't estimate
>>fiducial marks after the photo has been taken (i.e., with a 35 mm camera
>>with no marks on the lense).
>>
>>Like I said, sorry. This guy won't let it drop just by me telling him
>>it's impossible.
>>
>>PEg

t 08:36 PM 8/14/98 -0600, you wrote:
>The ortho function works with the required inputs! If he doesnt have the
>inputs for the function it is not our job to educate him one way or
>another. Just tell him there is a good reference in the the docs -
>
>See the following reference for more details:
>Wolf, R., 1974. Elements of Photogrammetry (2nd ed.), McGraw-Hill Inc., New
>York.
>
>Jim

*************************
Dear Peg,

Here's a post-honeymoon reply. The fiducials are used to create and affine
transformation (affine = 1-st order linear, with scaling, rotation,
translation and shear) that relates the x-y physical coordinates on the
camera focal plane (in mm usually) to the sample-line coordinates of the
scanned digital image. This is the interior orientation.

If one does not have fiducials and one still wants to orhto correct
imagery, one needs to "invent" fiducials. This is NOT recommended as you
know. Neither is going over Niagara in a barrel but some people seem to
like it.

We will not guarantee any results from such a process and you should tell
the guy this.

Nevertheless, here's what he could do:

1) find the exact size of the film frame he is using, from some definitive
source like Kodak,

2) identify the four corners, the four edge middles and the center of the
film frame, both in the camera and scanned image coordinates,

3) enter these in ENVI Interior Orientation as though they are from a valid
calibration report,

4) takee what he gets for results and be happy with it or drop it,

5) next time use a metric cmaer for metric camera applications.

It's like using a wrench to hammer in nails.

But yes he can fake his fiducials if he kwnos the size of the film and uses
the frame corners as de facto fiducials. I have no idea of how good the
results will be. Is this the same guy who is looking at Earth limb
pictures and wants Earth curvature in the solution? If so disregard all
the above and change your phone number.

Kindest of Regards,

Joe