While doing image classification, how many training sites (regions of interest) have to be taken from the study area?
Is there any method to do terrain correction in the hilly areas as shadows of the hills are showing as water bodies in the classified image?