I want to classify urban area from very high spatial resolution satellite images (Pan-sharpened image with 4 bands) in order to extract buildings only and to calculate the total built-up areas.
I am seeking to have very accurate results and overall classification accuracy must be above 85% - 90%. F1-score method must be used to test the algorithm performance which will be coded in Matlab. I will leave the selection of the classifier to your decision and you can combine different classifiers for achieving high accuracy. The classes which are required are:
Buildings, Vegetation, roads, other (shadow, water, etc).
An expert in this field is required.
So, please, if you find yourself is able to accomplish such a job, present some works in this scope. We will discuss everything.
Looking forward to hearing from you soon.