Spatial resolution improvement of digital images has significant applications in remote sensing and computer vision. Some former algorithms were （or supposed to be） almost for multiframe images of same spatial resolution and without geometric distortion. In this paper

a general method of interpolation and simulated sampling for spatial resolution improvement from multiframes of undersampled digital images is proposed based on the iterative metheds for signal reconstruction fain nonuniform sampling and the methods of projection onto convex sets （POCS）. The method can tackle multiframe images of various spatial resolutions

different image radiance

ralative geometric distortion

additive random noise

and so on. An idea of varying weights in iteration of parallel POCS methods is also discussed. Some experiments with resolution test pattern and multiangular remote sensing images show the convergence and effectiveness of the algorithms.