I calculated mean and S.d for all the bands of Landsat 8 and also mean and S.d of 8 indices. I also calculated GLCM and selected Contrast, Corr and dissimilarity of all 20 bands. While adding all the bands to the image, there are 120 bands in my image. Should I only use mean and s.d of all the bands and not the original ones? Because my classification shows error, is it because of more bands in my image?
P.S. I am using Google Earth Engine for the analysis.
Thank you
Hi,
You can use ACP to keep only the best bands and then you classify your image with
Simplifying your methods and results if possible is always a good way forward. If the features you are looking for can be identified using only a few bands then this would be a more straight-forward approach. This is true for all types of modelling scenarios, you want to be able to model reality as accurately as possible using as few variables as possible.
Please, select the number of bands as your theme of classification.
select the suitable bands for your classification and use training samples via google earth
- Badges
- Science topic
- Similar topics
- Data Mining
- Classification