Abstract: AFM images always have some background slope or curvature that must be removed from the image. Sources of the background can be an offset angle between the probe and surface, or curvature introduced into the image from the XYZ scanner. There are a number of background subtraction options that are possible. In this paper, we take use of mm-estimators, one most commonly employed robust regression technique to calculate the background in the image and enhance its speed via fast-s estimation approach. Numerical results prove its excellent performances in AFM image leveling compared with other traditional algorithms.

Key words:  AFM image leveling, robust regression, MM-estimator, fast-s estimation