A new method for multiframe image restoration and registration is presented. The
observations are sequences of low-resolution, undersampled, discrete frames. The result is
a restored high-resolution image. The restoration part is suitable for real-time
implementation since the computation consists of only a few complex operations.