Hi, I'm quite new to remote sensing softwares, though i have a backgound in the theories. I have these landsat images that i have to extract forest cover from, and convert them to vector polygons. That's my main objective.   I started in stacking my tiff bads into one image file, then deriving the ndvi using the ndvi tool (used the bands 3 and 4, red and near infrared). I did this to separate the vegetated areas from the non vegetated. Actually, I am stuck here. I believe that the flow of work would be something like this:   -create a mask to exclude the nuisance (clouds, cloud shadows, etc). -clssify the ndvi to delineate the different features on the ground -determine the reflectance value/threshold that should be used to identify which is forest cover and which is not -convert the class appointed as forest cover to vector   or something like that. Can anyone help me in this endeavor? Would there be a more better way in achieving the expected result? I hope you can guide me. thank you very much.