I am trying to classify a few google earth imageries of high resolution. The problem is that these images are only comprised of RGB band and NIR band is not available. While going for unsupervised classification I am not getting good results. Supervised classifications using 40 classes is not giving appealing results.
Interestingly few researchers from China have done the similar study to compare the capabilities of Google Earth Imageries (natural color combination) with QuickBird Satellite images (false color combination) and they got quite impressive results.
My study area is about 5 sq km, while their area was of 100+ sq km.
Does small area pose a problem in my case ?