The establishment of more accurate imaging of surface microstructures is needed. The most significant distortion in atomic force microscopy (AFM) imaging is induced by the probe tip shape, whenever the sample surface contains features whose dimensions are comparable to the probe tip size. The acquired AFM image is the dilation between the tip shape and the sample topography. To restore the original topographical profile, a numerical erosion procedure using a precise probe shape function is required. Here, a new technique for reconstruction of probe shape function using a well-defined nanostructure is proposed. First, AFM topography images of the given-shape nanostructure dispersed on flat substrates are taken. Then, a probe shape function is determined by a numerical calculation procedure. By using the experimentally determined probe shape function, the most probable surface morphologies from the observed AFM topography images of unknown samples can be extracted.