Non-Linear Restoration from a Single Frame Super Resolution Using Pearson Type VII Density

Super-resolution seeks to recover a high resolution image from one or more low resolution images. It is an ill-posed problem, with no consensus how best to devise image models that can both impose smoothness and preserve the edges in the image. Here we investigate the use of prior based on Pearson type VII density integrated with a Markov Random Field (MRF) model. We formulate two different versions, one that acts on the pixel level and another one that acts on the entire image. Having a single downsampled and noisy version of low resolution frame, we aim to obtain the high resolution image. We compare the state of the art of image priors in super resolution application and we discover that our image prior Pearson-MRF achieves the best performance in terms of qualitative measurement.