Efficient and Parallel Separable Dictionary Learning

Separable, or Kronecker product, dictionaries provide natural decompositions for 2D signals, such as images. In this paper, we describe an algorithm to learn such dictionaries which is highly parallelizable and which reaches sparse representations competitive with the previous state of the art dictionary learning algorithms from the literature. We highlight the performance of the proposed method to sparsely represent image data and for image denoising applications.