Image Extraction
ravi nooka
I am working in Remote Sensing image processing.
I have a satellite image, and want to extract sub-images (~1024X1024
pixels) from it. these small chips are used for matching with other
images.
My requirement is that these sub-images should have features like road
junctions, streams and man made things, so that these can be used to
match with other images.
So how to identify those featues in a images automatically.
regards,
ravi nooka.
Martin Leese
> Dear all,
> I am working in Remote Sensing image processing.
>
> I have a satellite image, and want to extract sub-images (~1024X1024
> pixels) from it. these small chips are used for matching with other
> images.
Why 1024x1024 ? Chips for matching remote
sensing images are typically *much* smaller
(usually about 1x1 km on the ground).
> My requirement is that these sub-images should have features like road
> junctions, streams and man made things, so that these can be used to
> match with other images.
> So how to identify those featues in a images automatically.
Typically you don't. You identify features
using an Eyeball Mark II (a trained
interpreter), and store the chip away along
with its accurate ground coordinates.
It sounds like you want to do things a
different way. Don't know of any research
specifically for this (perhaps other do).
The best I can suggest is that you search
the Science Citation Index and similar
places.
By the way, bends in streams and rivers tend
to move, so are less than optimal for what
you want. In some areas of the world (such
as rural Africa) you will be unable to find
any man-made objects, so will be reduced to
using them anyway. But where possible,
avoid streams and rivers. (Airport runways,
where they cross, are really great.)
--
Regards,
Martin Leese
E-mail: ple...@see.Web.for.e-mail.INVALID
Web: http://members.tripod.com/martin_leese/