| re-warping an image file [message #27937] |
Tue, 13 November 2001 14:22  |
glinka
Messages: 2
Registered: November 2001
|
Junior Member |
|
|
I have received Ikonos 4.9m imagery from an outside source, and the
corresponding mosaic overlayed horribly. I have divided up the one
image into five separate files and collected ground control points to
correspond with a Landat 7 scene. I am wondering if the warping that
was done initially will confound the re-warping of the image, and what
problems I can expect to encounter (if any). Is there possibly a
better way of doing this?
Hoping for any advice or direction,
Thanks,
Kristen
|
|
|
|
|
|
| Re: re-warping an image file [message #28042 is a reply to message #27937] |
Fri, 16 November 2001 06:15  |
cjengo
Messages: 28
Registered: June 1999
|
Junior Member |
|
|
Kristen,
It really depends on what has already been done to the images. The
geometry of Ikonos imagery is inherently more complex than Landsat.
Landsat is a line scanner, so terrain distortion is relatively easy to
fix. Ikonos is a push-broom that looks forward, backward, and
side-to-side so that it can collect multiple swaths of a study area
during a single overpass. And, because of its high resolution, it is
difficult in most places to find a DEM with high enough resolution to
do an accurate orthorectification.
We normally georectify Ikonos data using the rational functions
embedded in the NITF header. These describe (to fairly high
precision) the orbital model of the satellite during data collection,
and places the imagery pretty darn close to the right spot on the
ground. This also corrects for most geometry effects, except, of
course, for terrain distortion.
In some instances we've tried to tie Ikonos to Landsat. In short, it
doesn't work well. Terrain distortion in Ikonos can be any which way,
whereas it is consistent in Landsat. We also believe there are some
unresolved internal geometry artifacts caused by the optics.
How was your Ikonos data resampled? Single or multiple order
polynomial? Nearest neighbor or cubic convolution? If it was based
on an affine, warping again won't be too bad. If you do multiple
cubic type resamples, you'll blur the image considerably. Too many
nearest neighbor resamples, and it'll look awfully jagged. Best bet
would be to get ahold of the raw data, if possible, and do it
correctly from the start. Good luck!
Chris
glinka@rmc.ca (Kristen) wrote in message news:<9636b724.0111131422.3ee72bb4@posting.google.com>...
> I have received Ikonos 4.9m imagery from an outside source, and the
> corresponding mosaic overlayed horribly. I have divided up the one
> image into five separate files and collected ground control points to
> correspond with a Landat 7 scene. I am wondering if the warping that
> was done initially will confound the re-warping of the image, and what
> problems I can expect to encounter (if any). Is there possibly a
> better way of doing this?
>
> Hoping for any advice or direction,
> Thanks,
> Kristen
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
