Hello All,

How Does the (Inpho software) calculates the ground coordinates for each pixel in the image

I have a conceptual question regarding photogrammetry, what is the formula used to find the (X,Y,Z) on ground. All known software requires inputs like ( focal length, H of the plane, image exteriors (x,y,z, omega, phi, kappa), GCPs, pixel size) then the result is an orthophoto that have accurate coordinates, but I am curious on how does these software calculates the ground coordinate for each pixel in the image !

Is there any formulas/references on this, can I calculate a coordinate for one pixel manually just to check that the software is working fine or just to prove it for myself.

Best Regards

Ahmad

Hi Ahmad,

ok...this is getting extremely technical. Photogrammetry uses a lot of formulas. You are correct concerning the inputs:

- interior orientation (inner geometry of the image, i.e. all the camera parameters like sensor size, focal length, principal point location, pixel size, image rotation...) plus distortion parameters for corrections

- absolute orientation (position and rotation xyz, omega phi kappa), derived from bundle adjustments with GCP or GNSS/IMU etc.

from these parameters, you can take an image pixel, and calculate the position where the ray through the camera lense would intersect the terrain. Of course you need to have a terrain model . The computation (also considering atmospheric effects for corrections and earth's curvature effects...) is quite complex.

there is an article http://www.isprs.org/proceedings/XXXIII/congress/part4/400_XXXIII-part4.pdf that might give you some initial information. Other than that, there is quite a lot of literature available https://books.google.de/books?id=L5QcP5QkUm4C&pg=PA410&hl=de&source=gbs_toc_r&cad=3#v=onepage&q&f=false (this one, however is in german...).

You also need to consider that the ortho-pixel itself is a resampled pixel ( the orthophoto has a constant scale, while the aerial image itself has different scale depending on the object distance (DSM) that is varying...) - so there is not a 1:1 ratio of orthopixels to aerial image pixels.

I hope this helps for a first answer... let me know if you need more details - but then it will really get complex :-)