As an important component of Atomic Force Microscopes (AFM), a piezo-scanner exhibits some undesired nonlinear characteristics, among which the inherent hysteresis largely decreases positioning accuracy during scanning and nano-manipulation process. To alleviate this problem, an image-based approach is proposed in this paper to model and then compensate for the hysteresis behavior of the piezo-scanner. Specifically, some scanning images over standard samples are utilized to identify the parameters of the classical Preisach model (CPM) of hysteresis. On the basis of the obtained model, an inversion-based technique is adopted to design a compensator for the hysteresis of the piezo-scanner. The proposed algorithm presents such advantages as low cost and little complexity since no nanoscale position sensor is required to collect identification data. Some scanning and nano-imprinting results are included to demonstrate the performance of the proposed strategy.