Sparse Representations and Dictionary Learning: from Image Fusion to Motion Estimation