We present a quantitative method of image analysis that automatically corrects for system drift in comparisons of images before and after an event under study. The method utilizes two-dimensional correlations followed by automatic z scaling, shifting, and the calculation of a difference image from the before and after images. Optical reflectance near-field scanning optical microscope (NSOM) images that show the effects of oxygen electromigration are used as an example system. The tunnel current that induces the atomic motion is provided by the metallic NSOM aperture in a scanning tunneling microscope mode. The analyzed (difference) image clearly shows the effects that are otherwise obscured by native oxygen concentration variations, demonstrating the utility of the method. The related issues of false positive/negative probabilities for given noise levels and thresholds are also discussed. (C) 2003 American Vacuum Society.