Image enlargement by localized sampling

Changing resolution of images is a common operation. We present a novel image interpolation algorithm and apply it to image enlargement. The algorithm is based on localized sampling which develops Shannon sampling in the sense that it uses only local samples to reconstruct a signal. The method achieves exponentially decaying accuracy of approximation, and provides rigorous means for obtaining any desired accuracy for sampling. For the quality of interpolated image, approximation errors show that our method generates better results than existing methods used in common commercial software. These features give it the potential to be used in image processing applications.