Research on filtering of DEM data based on the Fisher functions

Random Error in intensive DEM data brings noise and interference into net model. The filtering method is proposed that based on the Fisher functions to filter the random noise existed in the intensive DEM, and with the peer group members of the Gaussian weighted average values to replace the average values of the entire local window, Experiments prove that step-divided filtering can not only effectively eliminate noise, improve DEM precision, but better reflect hypsography of the whole experimental area without missing any detailed terrain features.