A method of post-acquisition upsampling for scanning x-ray microscopy is developed in this study to improve the spatial resolution beyond the Nyquist frequency as determined by the intervals of a raster scan grid. The proposed method is applicable only when the probe beam size is not negligibly small compared with the pixels that constitute a raster micrograph-the Voronoi cells of a scan grid. The unconvoluted spatial variation in a photoresponse is estimated by solving a stochastic inverse problem at a higher resolution than that at which the data are acquired. This is followed by a rise in the spatial cutoff frequency due to a reduction in the noise floor. The practicability of the proposed method was verified by applying it to raster micrographs of x-ray absorption in Nd-Fe-B sintered magnets. The improvement thus achieved in spatial resolution was numerically demonstrated via spectral analysis by using the discrete Fourier transform. The authors also argue for a reasonable decimation scheme for the spatial sampling interval in relation to an ill-posed inverse problem and aliasing. The computer-assisted enhancement in the viability of scanning x-ray magnetic circular dichroism microscopy was illustrated by visualizing magnetic field-induced changes in domain patterns of the Nd2Fe14B main-phase.