empty pixels occurs after classification - in arcmap you can use Generalization tools such as majority filter. However is there a script you could write to better fix it? for example - some of the tools are bad at fixing empty pixels on borders.
Ok to be more specific. Is it possible to use a python script to select empty pixels and reclassify them based on distance to neighbours? The nibble tool selects No Data pixels and reclassifies them according to distance to neighbours. But I have a question asking if there is a way to better do it - as with current methods often areas such as borders are missed out (due to lack of neighbours)
This is for Arcmap10
