In this paper, a new algorithm is proposed based on coupled dictionary learning with mapping function for the\nproblem of single-image super-resolution. Dictionaries are designed for a set of clustered data. Data is classified into\ndirectional clusters by correlation criterion. The training data is structured into nine clusters based on correlation\nbetween the data patches and already developed directional templates. The invariance of the sparse representations\nis assumed for the task of super-resolution. For each cluster, a pair of high-resolution and low-resolution dictionaries\nare designed along with their mapping functions. This coupled dictionary learning with a mapping function helps in\nstrengthening the invariance of sparse representation coefficients for different resolution levels. During the\nreconstruction phase, for a given low-resolution patch a set of directional clustered dictionaries are used, and the\ncluster is selected which gives the least sparse representation error. Then, a pair of dictionaries with mapping\nfunctions of that cluster are used for the high-resolution patch approximation. The proposed algorithm is compared\nwith earlier work including the currently top-ranked super-resolution algorithm. By the proposed mechanism, the\nrecovery of directional fine features becomes prominent.
Loading....