I need to classify LISS IV data for land use planning but there is a lot of mixing signatures in between Roads and riverbed, Agriculture and Forest. So, i need some Hybrid Classification methods to do the same.....
You can use some indexes to identify some mixed classes. Later, you can mask succesfull classes and reclassify the remaining part of the image.
You can also apply a clustering algorithm to identify some classes and as I sugegst abıve you can mask them and continue with a supervised approach.
The following link may be useful for you, please check
http://research.ijcaonline.org/volume119/number8/pxc3903779.pdf
http://www.isprs.org/proceedings/XXXVI/part4/RS-F-6.pdf
https://www.iirs.gov.in/iirs/sites/default/files/StudentThesis/dutta_amitava_iirsmsc.pdf
Hello, LISS IV image contains less spectral information, I would suggest to derive local specific knowledge based rules... the follwing paper (open access ) might be helpful to you to derive such simple and efficient rules.
http://www.sciencedirect.com/science/article/pii/S1110982315000381
All the best
You could try spectral unmixing or sub-pixel classification as this technique will give you 'within a pixel' estimate or proportions of a given class.
- Badges
- Science topic
- Similar topics
- Instrumentation Engineering
- Instruments