In imaging a sample using an electron microscope, in order to reduce a time for focusing, a scanning range of a Z coordinate is reduced to complete focusing by obtaining SEM images such that: for a first predetermined number of portions, focal positions of an electron beam in obtaining each SEM image are moved in a predetermined range; then, a curved surface shape of the surface of the sample is estimated by using information relating to the focal positions of the electron beam in the first predetermined number of portions; after the images are taken, the range in which the focal positions of the electron beam are moved for scanning the electron beam on the surface of the sample is made to be narrower than the predetermined range by using the curved surface information estimated, thereby performing scanning to take the images of the sample.