hello,
i am working on a land use classification in the Mekong delta, based on
quickbird data using Grass6.4 RC5 and have 2 questions.
The scene is in parts cloudy, especially my training area. is there an
option to do some correction to get rid of the clouds?
i already did a classification with 8 classes (about 90 sample) but i'm
not really satisfied with the result because its so inhomogeneous with
many small fractions. Is there some literature or experience how to
classify rural areas with small huts, rice paddys (mature, fallow, under
water) and mixed orchards?
thanks
christian
Christian Röttger schrieb:
hello,
i am working on a land use classification in the Mekong delta, based on
quickbird data using Grass6.4 RC5 and have 2 questions.
The scene is in parts cloudy, especially my training area. is there an
option to do some correction to get rid of the clouds?
i already did a classification with 8 classes (about 90 sample) but i'm
not really satisfied with the result because its so inhomogeneous with
many small fractions. Is there some literature or experience how to
classify rural areas with small huts, rice paddys (mature, fallow, under
water) and mixed orchards?
sorry, i forgot to say which modules i used. first i did a unsupervised
cluster classification and a false colour composit plus gps ground
points to generate my training areas. after that i tried i.maxlik and
then i.smap to get a better result but that also didn't satisfy me.
i will also try some modification parameters, or generate more training
areas.
regards
christian
Hi Christian,
Well, I don't think you can do anything about the clouds. You will
just have to live with that, or work with radar images, which is an
entirely different beast...
As for the classification, if you want to remove speckle (small huts
for instance), maybe a median filter could help.
On the rice paddies, if they are very different in your image, I'd
create separate classes for each rice paddy type (mature, fallow, etc)
and regroup them latter.
Good luck
Daniel
On Tue, Nov 17, 2009 at 5:37 PM, Christian Röttger
<chris.roettger@uni-muenster.de> wrote:
hello,
i am working on a land use classification in the Mekong delta, based on
quickbird data using Grass6.4 RC5 and have 2 questions.
The scene is in parts cloudy, especially my training area. is there an
option to do some correction to get rid of the clouds?
i already did a classification with 8 classes (about 90 sample) but i'm
not really satisfied with the result because its so inhomogeneous with
many small fractions. Is there some literature or experience how to
classify rural areas with small huts, rice paddys (mature, fallow, under
water) and mixed orchards?
thanks
christian
_______________________________________________
grass-user mailing list
grass-user@lists.osgeo.org
http://lists.osgeo.org/mailman/listinfo/grass-user