Image Sparse Representation Based on a Nonparametric Bayesian Model
In recent years there has been a growing interest in the research of image sparse representation. Sparse representation based on over-complete dictionary become another hot topic in the field of image processing. In this paper a Nonparametric Bayesian model based on hierarchical Bayesian theory is proposed. In this model a sparse spike-slab prior is imposed on sparse coefficients and the Non-parametric Bayesian techniques based on sparse image representation are considering for learning dictionary. Proposed model can learn an over-complete dictionary from original image. Furthermore, the unknown noise variance can be estimated from noisy image. As regards to the image sparse representation, proposed model obtains good sparse solution. Comparing to other state-of-the-art image sparse representation method, this model obtains better reconstruction effects.
X. H. Ding and X. B. Chen, "Image Sparse Representation Based on a Nonparametric Bayesian Model", Applied Mechanics and Materials, Vol. 103, pp. 109-114, 2012