Quantitative detection of defects in atomic structures is of great significance to evaluating product quality and exploring quality improvement process. In this study, a Fourier transform filtered sampling Moiré technique was proposed to visualize and detect defects in atomic arrays in a large field of view. Defect distributions, defect numbers and defect densities could be visually and quantitatively determined from a single atomic structure image at low cost. The effectiveness of the proposed technique was verified from numerical simulations. As an application, the dislocation distributions in a GaN/AlGaN atomic structure in two directions were magnified and displayed in Moiré phase maps, and defect locations and densities were detected automatically. The proposed technique is able to provide valuable references to material scientists and engineers by checking the effect of various treatments for defect reduction.