Hi, david...Thank you so much to share so informative and easy to learn lecture. David i have problem to use the Region group tool in generalization menu. i want to use it in unsupervised classification for remove some speckles. but i am not sure how to select the number of pixel values in the image and what criteria in which i should use that these values could have been selected. and further this output can be use in nibble tool for removing those speckles.
Thanks for the tutorial! One question, how do you calculate the amount of points that you need to check in order to have a robust kappa test? You used 100 points to each class in the example. Why? Is there a rule for that? Thank you so much!
thanks a million, very very helpful, but I was wondering if the two maps area raster (they are both landuse maps one original and one predicted) how can I do the same thing?
Missy Skeeter and if we have not ground truth polygon rather we have google earth land cover name that have been generated by surfing the google earth using kml file of those points that have generated using random points.
If you have 100 polygons the ArcGIS Crate Random Points geoprocessing tool would create 100 random points inside each polygon resulting in 100 X 100 = 10,000 total random points
To create random locations, use the geoprocessing tool Create Random Points setting the extent to the same extent as your classified raster, and specifying the number of random points desired.