BTF Potts compound texture model

Abstract

This paper introduces a method for modeling mosaic-like textures using a multispectral parametric Bidirectional Texture Function (BTF) compound Markov random field model (CMRF). The primary purpose of our synthetic texture approach is to reproduce, compress, and enlarge a given measured texture image so that ideally both natural and synthetic texture will be visually indiscernible, but the model can be easily applied for BFT material editing. The CMRF model consist of several sub-models each having different characteristics along with an underlying structure model which controls transitions between these sub models. The proposed model uses the Potts random field for distributing local texture models in the form of analytically solvable wide-sense BTF Markovian representation for single regions among the fields of a mosaic approximated by the Voronoi diagram. The control field of the BTF-CMRF is generated by the Potts random field model build on top of the adjacency graph of a measured mosaic. The compound random field synthesis combines the modified fast Swendsen- Wang Markov Chain Monte Carlo sampling of the hierarchical Potts MRF part with the fast and analytical synthesis of single regional BTF MRFs. The local texture regions (not necessarily continuous) are represented by an analytical BTF model which consists of single factors modeled by the adaptive 3D causal auto-regressive (3DCAR) random field model which can be analytically estimated as well as synthesized. The visual quality of the resulting complex synthetic textures generally surpasses the outputs of the previously published simpler non-compound BTF-MRF models.