Super-Resolution Using Manifold Learning
Dattatray L. Karambelkar, P.J. Kulkarni · 2011
In this paper, a novel method for solving single image super-resolution problem is proposed. Objective is to recover a high resolution version of the given low resolution image. This method takes into account, a popular method of dimensionality reduction. It is believed that, small image patches in the low-resolution and high-resolution images form manifolds with similar local geometry in two distinct feature spaces. In locally linear embedding, focus is on reconstruction of a feature vector for a given patch using its neighbors in the feature space. This feature vector is used for the super resolution.