The invention provides an electromagnetic imaging super-resolution method based on LR-geometric interpolation. In an electromagnetic imaging system, image fuzziness is caused by filtering of a diffraction limited system and undersampling. For a problem that image spatial resolution can not be enhanced in a routine electromagnetic imaging super-resolution algorithm in the prior art, a super-resolution algorithm combining average geometric interpolation and LR iteration. The method specifically comprises steps that firstly, average geometric interpolation of a degraded image is carried out, image pixel points are enhanced, the image information is enhanced, secondly, an LR iteration super-resolution algorithm is employed to carry out super-resolution recovery of a system of a known point diffusion function, compared with a traditional algorithm, image spatial resolutions after recovery under noise and no-noise conditions are respectively enhanced by 70% and 20%, moreover, the inhibition effect on external noise is realized to a certain degree, and correctness of the method is proved through simulation and experiments.