Image restoration using resolution expansion is important in many areas of image processing. This paper introduces a restoration method for low-resolution text images which produces expanded images with improved definition. This technique creates a strongly bimodal image with smooth regions in both the foreground and background, while allowing for sharp discontinuities at the edges. The restored image, which is constrained by the given low-resolution image, is generated by iteratively solving a nonlinear optimization problem. Lowresolution text images restored using this technique are shown to be both quantitatively and qualitatively superior to images expanded using the standard methods of linear interpolation and cubic spline expansion. Experimental results demonstrate that text images created by this new algorithm improve optical character recognition accuracy more than images obtained by existing expansion methods.