This paper is devoted to a new variational model for color texture modeling and color image decomposition into cartoon and texture. A given image f in the RGB space is decomposed into a cartoon part and a texture part. The cartoon part is modeled by the space of vector-valued functions of bounded variation, while the texture or noise part is modeled by a space of oscillatory functions, dual in some sense of the BV space. Examples for color image decomposition, color image denoising, and color texture discrimination and segmentation will be presented.
Luminita A. Vese, Stanley Osher