Although majority
filter is easy to implement, its weaknesses are apparent. First, the
size of neighborhood for the filter has to be very large for noise to
be sufficiently removed, while a large size neighborhood may
alter the boundaries between classes and create zigzag bounding
polygons. Moreover, a loss of meaningful information in
classified data is shown because of the geometric and dimensional
non-correspondence of real objects with the moving window
implementation matrix. Finally, such methods need extensive
editing operations on classified images in order to be stored in
GIS databases