A UNIFIED TEXTURE MODEL-BASED ON A 2-D WOLD-LIKE DECOMPOSITION

The paper presents a unified texture model which is applicable to a wi de variety of texture types found in natural images. This model leads to the derivation or texture analysis and synthesis algorithms designe d to estimate the texture parameters and to reconstruct the original t exture field from these parameters. The model is highly motivated by f indings about human vision. The texture field is assumed to be a reali zation of a regular homogeneous random field, which is characterized i n general by a mixed spectral distribution. On the basis of a two-dime nsional (2-D) Wold-like decomposition for homogeneous random fields, t he texture field is decomposed into a sum of two mutually orthogonal c omponents: a purely indeterministic component and a deterministic comp onent. The deterministic component is further orthogonally decomposed into a harmonic component, and a generalized-evanescent component. The purely indeterministic component is represented by a 2-D, nonsymmetri cal-half-plane, finite support AR model. The harmonic random field is a sum of 2-D harmonic components of random amplitude and phase. The ge neralized evanescent field consists of a countable number of wave syst ems all traveling in directions of rational tangent, and all modulated by 1-D purely indeterministic processes in the orthogonal dimension. Both analytical and experimental results show that the deterministic c omponents should be parametrized separately from the purely indetermin istic component. The model is very efficient in terms of the number of parameters required to faithfully represent textures. Reconstructed t extures are practically indistinguishable from the originals.