ABSTRACT
ABSTRACT: Regression methods to reduce image distortions of a three-dimensional scanning electron microscope equipped with two pairs of secondary electron detectors have been studied. Such distortions originated in the electron beam scanning cause errors in the measured heights of the surface topographies of specimens. These errors, which decrease rapidly at higher magnifications, are quantitatively evaluated and in order to remove them effectively, some regression methods are examined from the viewpoints of the amplitude transmission characteristics and the processing time. Parabolic regression followed by Spline filtering is proposed as the most effective method. Notice for applying this method is also discussed.
