Imaging techniques often suffer from distortion effects. Former methods of reducing these distortions have been based either on improving the imaging technique (i.e., to avoid distortions) or on the use of reference samples (i.e., to determine the distortion field by imaging of a known structure. We present a novel method of correcting image distortion by evaluating the imaged position changes due to two small sample position shifts. The algorithm allows us to calculate a vector field, which enables us to determine the “undistorted” position of any point of the image. The presented method has very low presuppositions about the sample, requires no reference samples, and is applicable to any type of image distortion. In addition to the presentation of the method's theoretical basis and a description of the computational method, we present corrected secondary ion mass spectroscopy (SIMS) images of a regular structure (a copper grid) as well as a stochastic distribution (sodium impurities) to show the results of empirical data.