A wavelet and PDE method for optical image interpolation

Image interpolation occurs anytime you resize or remap (distort) your image from one pixel grid to another, it works by using known data to estimate values at unknown points. In this paper, aiming at improving the sharpness and removing the sawtooth at edges, a novel method based on integration of wavelet and PDE is proposed. In order to get an initial magnification result, wavelet interpolation method is used to resample the image. After that, a so-called self-snake PDE model is introduced in post-processing of the interpolated images. In this paper, we compare the results of this method with other popular methods. Edge sharpness and structure similarity are used as a quantitative measure of images´ quality. The experiments show that our method based on wavelet and PDE has a better performance than others.