PROJECT TITLE :

Non-local feature back-projection for image super-resolution

ABSTRACT:

Image super-resolution (SR) for a single low-resolution image is a crucial and challenging task in Image Processing. During this study, the authors propose a unique non-local feature back-projection method for image SR, which can effectively scale back jaggy and ringing artefacts common, in general, iterative back-projection (IBP) methodology. In their technique, the target high-resolution (HR) image is obtained by projecting reconstructed errors back to HR image iteratively. To optimise the initial HR image and constrain anisotropic errors propagation during IBP method, an efficient non-native feature interpolation algorithm is designed. Specially, edge data is employed as constraints to form the interpolation surface preserve higher shape. Furthermore, as post-processing, non-local similarities are utilised to get rid of noise and irregularities induced by errors propagation. Experimental results show that their methodology achieves higher performance than state-of-the-art ways in terms of each quantitative metrics and visual qualities.