3-D texture characterization based on Wold decomposition and higher order statistics

This paper deals with the parametric 3D models for volumes and image blocks based on the so-called Wold decomposition theory. This decomposition raises purely indeterministic, harmonic and evanescent components. Based on this theory, we recently proposed new explicit parametric models able to describe both the spectral support and spatial structure of each component. When the purely indeterministic component is assumed Gaussian and the evanescent component vanishes, we state the problem of separating the 3D texture into a deterministic component and an indeterministic one such as a 3D harmonics retrieval problem in a noisy background. We propose to achieve such a task using a new 3D spectral method based on higher order statistics (HOS).