Re: Measuring angles in an image.



Say, I have taken the image of certain area covering 50 meters
square. My principle point being a building top corner , I need
to get
the angle it makes with a target object ,say a flag pole top
with respect to principle axis . To locate the point I will need to
zoom until the pixels pops as a means to get
the accuracy . Does that disturb the spatial calibration ? Remember
that I am working on image that is already captured.
If I don't zoom the image I am left with poor accuracy . Zooming
often arises a question in my mind ( Does "zooming in" in
common image viewers means that you are going beyond 100% , that
again creates the problem because if the application is performing a
sort of rendering ( digital zoom) I loose original contents of my
image and what I focus may resemble nothing in real world ).

Regarding spatial calibration , I work the real distance between two
image elements , note how many pixels it occupies and then I can
easily work distances on other photographs too if they were taken
at same distance as former one and that same camera was used
( excluding the lens distortions and environmental parameters ) .
But
the distances are only correct if they lie in the same plane of the
reference object . For example, if my flag pole is far behind the
building , on image they have small spatial separation but in
reality they are far from each other. This gives a thought of DN,
value of the elements , can that be incorporated to find the real
distances.

Regards ,

Geobird
.



Relevant Pages

  • Re: Measuring angles in an image.
    ... image  you are going to capture (principle point), ... to camera axis. ... How do you  think of distances? ... such as taking into account field curavature, distortion, etc. Go to ...
    (sci.image.processing)
  • Re: The return of "DC in 2011"
    ... distances. ... It could, in principle, be in the same city ... Lt. Hiroo Onoda of that war. ... Denver's city bus system relenting on demanding ID of riders, ...
    (rec.arts.sf.fandom)

Quantcast