The spatial image interpolation problem can be stated in its most general form as follows: Given a digital image obtained by sampling an ideal continuous image on a certain spatial grid, obtain the pixel values sampled on a different spatial grid. If the new sampling grid is denser than the input sampling grid, then the interpolated image has more samples than the input image. This case is referred to as zoom-in and the main challenge is to compute the values of the new pixels with high precision without creating artifacts. If the new sampling grid is sparser than the input sampling grid, then the interpolated image has fewer samples. This is referred to as zoom-out. Since zoom-out effectively performs down-sampling, the challenge is to obtain a smaller image free of aliasing artifacts with minimal blur.