|
|
| GeoCommunity Mailing List |
| |
| Mailing List Archives |
| Subject: | RE: GISList: orthorectifying vs. triangulated warping |
| Date: |
08/30/2002 08:41:52 AM |
| From: |
Tomaz Skrbinsek |
|
|
I found a really detailed article on that topic:
http://www.gisdevelopment.net/aars/acrs/2000/ts4/digi0012.shtml
Don't you just love Google? :)
Cheers,
Tomaz.
-----------------------------------------------------------------------
bine@kibla.org Sometimes I think the surest sign
that intelligent life exists elsewhere=20
Tomaz Skrbinsek is the fact that none of it has tried to contact us
-----------------------------------------------------------------------
> -----Original Message-----
> From: Iv=E1n Alfaro [mailto:alfasig@ceibo.entelnet.bo]
> Sent: Friday, August 30, 2002 1:52 AM
> To: gislist@geocomm.com
> Subject: RE: GISList: orthorectifying vs. triangulated warping
>=20
> I guess Tomaz has explained the problem perfectly. I would like to put
> something else in the table: What happens when you want to operate
with
> a high resolution satellite image, for example Ikonos (4m) or Quick
Bird
> (2.5m). Does the concept of "close to infinte" applies?, or due to the
> high detailed elements you can appreciate, you really need to use a
DTM
> and perform orthorectification?
> Thank you,
>=20
> Iv=E1n Alfaro
>=20
> -----Mensaje original-----
> De: Tomaz Skrbinsek [mailto:bine@kibla.org]
> Enviado el: Viernes, 30 de Agosto de 2002 06:30 a.m.
> Para: 'Viktoras Didziulis'
> CC: gislist@geocomm.com
> Asunto: RE: GISList: orthorectifying vs. triangulated warping
>=20
> It all depends on your terrain and your image. The problem is radial
> deformation: When you make an aerial photo, your 3D terrain gets
> projected on a 2D surface. This is ok in the center of your
photograph,
> as the light reflected from the terrain is perpendicular to both the
> photograph and the terrain. The problem occurs toward the edges, since
> the points that have higher elevation get projected outward relative
to
> the points with lower elevation, which get projected inward. For
> example, imagine looking at a tall industrial chimney from a hill. The
> base of the chimney seems where it should be, on the ground. The top
of
> the chimney, if you imagine this as a 2D photo, lies on the fields far
> beyond the base. The same thing happens to the terrain.
>=20
> So, the radial deformation depends on the altitude of the camera
(lower
> the altitude, higher the deformation), differences in terrain
elevations
> (higher the differences, higher the deformation) and view angle of the
> camera (wider the angle, higher the deformation toward the edges).
>=20
> So, ideally, if your terrain was flat as a pancake, or if the
photograph
> was taken from infinite distance, ortorectifying would do exactly the
> same as warping. The difference between the both procedures is that
> ortorectification takes into account the radial deformation and
corrects
> it (so, you will need a DEM of the area you want to ortorectify,
camera
> flight altitude and camera lens parameters), while warping just does a
> rubber sheet warp based on a limited amount of reference points.
>=20
> It all depends on your image and your application: if you have a
> satellite photo, than you get quite close to having your camera at an
> infinite altitude above the terrain, so the warping routine actually
> does a pretty good job. However, if you have an aerial photograph
taken
> from an airplane, the radial deformation really kicks in and you can
> pretty much forget warping if you want to have any precision in your
> final image. Still, if your terrain is really flat, you might get a
> fairly decent result. It may also work if you need your image only for
> presentation purposes and you're not really concerned about precision.
>=20
> You can read more about the topic in practically any book on basic
> remote sensing. I apologize if the terminology I used in this
> explanation is incorrect, since I've studied the subject in another
> language.
>=20
> Regards,
>=20
> Tomaz.
>=20
>
-----------------------------------------------------------------------
> bine@kibla.org Sometimes I think the surest sign
> that intelligent life exists elsewhere
> Tomaz Skrbinsek is the fact that none of it has tried to contact us
>
-----------------------------------------------------------------------
> > -----Original Message-----
> > From: Viktoras Didziulis [mailto:firekv@hotmail.com]
> > Sent: Friday, August 30, 2002 7:51 AM
> > To: gislist@geocomm.com
> > Subject: GISList: orthorectifying vs. triangulated warping
> >
> > Actualy is there really a big difference in an outcomming result
using
> a
> > "real" orthorectifying routine and image warping with triangulation
> > algorithm having dense grid
|
|
Sponsored by:
For information
regarding
advertising rates
Click Here!
|
