Selective Sparse Coding Based Coupled Dictionary Learning Algorithm for Single Image Super-Resolution

In this paper a new strategy of multiple dictionary learning is proposed for the problem of super-resolution. A two way clustering mechanism is proposed for classification. Dictionaries are obtained for each cluster by coupled dictionary learning with mapping functions. Clustering of training data is carried out by using two approximate scale invariant features. This is followed by coupled dictionary and mapping learning which further helps in making the sparse representation invariant to resolution blur. This mechanism provides a selective sparse coding over multiple dictionaries. At the reconstruction phase each patch is recovered by selective sparse coding and dictionary learning. Experiment results indicate that the proposed algorithm is on par with existing state-of-the-art algorithms. The proposed algorithm is able to recover directional features more accurately.