Hi All, I want to classify 5 Landsat images together, of the same path/row but of different date. What I used to do is to stack the images together and the result in this case is one images with 30 bands (6 bands per image). But in order to classify them all I have normally to apply one mask. What I am wondering is… if it possible to apply different masks (0 or 255 value to particular pixels) to different images, and then to classify them all together. In that case I am thinking that is necessary to have an algorithm in ENVI that ignores particular pixel values (eg. 0 or 255) and classify those pixel using the other bands only. If this is feasible, one pixel will be classified using 18 bands, while other 24 and etc. Best Regards, Spyros