Atomic Force Microscope (AFM) is an ultra-precision instrument for characterizing nano-scale morphological structures, and the morphology of its probe tip directly affects the quality of scanned images and the measurement accuracy of critical dimensions. To accurately characterize the three-dimensional information of AFM probe tips, research on nano-scale characterizers and AFM probe tip reconstruction algorithms is carried out. Firstly, a multi-layer film deposition technique is studied, and a scheme of converting film thickness into critical dimension line-width through multiple steps of film growth is proposed to develop nano-scale characterizers. Secondly, aiming at the broadening effect occurring in the process of AFM probe measuring nano-scale line-widths, the blind reconstruction algorithm for probe tips is studied, and the probe reconstruction process and critical dimension measurement analysis are carried out. Finally, an AFM probe reconstruction algorithm based on scanning trajectory is proposed. This algorithm combines nano-scale characterizers and TEM (Transmission Electron Microscope) testing technology, which can accurately reconstruct the three-dimensional information of probe tips and reduce the noise of scanned images. Two algorithms are used to measure the critical dimension line-width structures. The results show that the blind reconstruction algorithm can reduce the influence of the broadening effect on line-width measurement to a certain extent. In contrast, the measurement accuracy of the AFM probe reconstruction algorithm based on scanning trajectory reaches 0.5 nm, which can meet the measuring requirements of nano-scale line-width structures.