Image decomposition via learning the morphological diversity

Image decomposition aims to separate different features in images. Based on dictionary learning (DL) techniques, this letter discusses two new algorithms for image decomposing into a linear combination of morphological components. The proposed algorithms can be viewed as the extensions of DL-based image denoising algorithm. Experiments show that the learned dictionaries by the proposed algorithms can describe the different components of image effectively and leads to high quality image decomposition performance.

► We extend DL-based image denoising technique to image decomposition.
► We present two algorithms to solve proposed model.
► The learned dictionaries can describe the different components of image effectively.
► The proposed algorithms achieve high quality image decomposition results.