Hello!

If you are using supervised classification, as a first step you can use visual interpretation techniques to collect the training datasets.

Just check the tutorials on http://dst-iget.in/tutorials/iget-rs/iget-rs-001 portal.

If I collect training data set in this manner then after classification how I will check or create confusion matrix.

I recommend you to check out this book:

http://www.univpgri-palembang.ac.id/perpus-fkip/Perpustakaan/Geography/Metodologi/ebooksclub.org__Field_Methods_in_Remote_Sensing.pdf

Good luck!

For data testing, you can use Quantum gis.

Ground check data is the best for data training. After that, you can use high-resolution satellite image from Google Earth.

Here our work in Sumatra, Indonesia

http://onlinelibrary.wiley.com/doi/10.1111/conl.12256/epdf

For training data you should possibly go to the field with GPS tools and collect ground truth data plus the XY coordinates each sample point.In addition if available you may use aerial photograph or topo maps,Google Earth as supplemental to the ground truth information. Then encode the GT data with their respective XY coordinate to be able to analyse confusion matrix of Classified image and Ground truth data.

For Accuracy assessment of classified image, here is a nice piece of paper:

http://www.sciencedirect.com/science/article/pii/003442579190048B

https://www.rit.edu/cos/colorscience/mellon/software.php

check multispectraltools in matlab

Hi all

Please I need multispectral images in food , fruits , skin diseases (more than 25 images)

thank you.

Hi .... I am working in LWIR and need Multispectral data set in this region.?