Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

I am working with multiple aerial images. My idea is to model 3d objects (only upper parts). I am having known orientation parameters. As I am new to this field so that, I want to clarify few general things.

Question1 If I work with a small area covering a given aerial photo (say 1:8000 scale), then I think I should select aerial photos which contain my desired area in the middle (option1) rather than at the edges (like option2, 3, 4 and 5). see the figure.

Then, I feel, the ex-tractable objects are less distorted if I select photographs which cover my area in the middle than objects near to the edges of the photograph. Does this correct idea and does this always must? Please comment me. enter image description here

Question2 My second question is, as I am working with multiple images and my intention is to intersect viewing planes (the plane forming by viewing rays) relevant to corresponding feature lines (I have extracted them) and then to get 3d information; then

Should I select at least nadir view (i.e. rays coming from O1 - projection centre) and few other photographs (like O2) which make large angle with that (O1) plane? (As I feel, plane near to nadir point is more accurate and on the other hand it needs large angles to get accurate intersection than what is given by O3-O4) Do you think, if I follow this way, can I estimate accurate 3D information. see figure enter image description here I think, this logic implies that I should always select planes which make large angle rather than small angle like (O3-O4)

So, I need your comments whether the logic that I am using is correct or not. ARE there any suggestions when making 3d information through viewing plane intersection with multiple images?

Thanks in advance

share|cite|improve this question
In real life photography, features closer to the rim of the picture are subject to deformations not modeled by projective transformations. So the closer to the center you are, the more accurate and useful a simple math approach will be. Otherwise you might have to correct for these kinds of errors, like e.g. Hugin does with lens correction. – MvG Jun 5 '13 at 13:51
@MvG: ok. thank you.. If I am neglecting these errors and if I am considering mainly the viewing ray geometry, then what could you say about my thinking... can I select photographs where they make large angles for a particular object... likewise, can I select different multiple photographs from the given list to get more accurate result. Please comment on this question as I was thinking many days on this.. thank you again. – gnp Jun 5 '13 at 14:01
@MvG: could you please give some idea on my second question as well please. – gnp Jun 5 '13 at 14:10
I don't have any real experience there. I guess that pictures taken vom very different angles could make matching up the features rather difficult. But assuming your features are sufficiently point-like, the larger the angle the larger the accuracy I'd expect. – MvG Jun 5 '13 at 17:17

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Browse other questions tagged or ask your own question.