PROJECT TITLE :

A Non-Local Dual-Domain Approach to Cartoon and Texture Decompositioní_

ABSTRACT:

Cartoon and texture decomposition is the subject of this article. As microtextures are characterised by their power spectrum, the information offered by the power spectrum of image patches can be used to extract cartoon and texture components There are statistically significant spectral components in the spectrum of a patch that contribute to the spectrum of a non-textured patch in comparison to a null hypothesis. Based on a simplified cartoon depiction of the literature, the null hypothesis model is constructed. When we talk about "dual domain," we mean that both spatial and spectral decompositions are used as inputs to our suggested spectral technique. Power spectrum of patches with similar textures is also used to build the statistical model. The proposed method is therefore a non-local technique. There are numerous examples of how this technique can be used in various fields, including fine art painting and remote sensing imaging.