A Nonparametric Multiscale Markov Random Field Model For Synthesizing Natural Textures
Rupert Paget, D. Longstaff · Information Sciences, Signal Processing and their Applications · 1996
In this paper we present a non-causal, non-parametric, multiscale, Markov random field (MRF) texture model. The model is capable of capturing the characteristics of and synthesising a wide variety of textures, varying from the highly structured to the stochastic. We introduce a novel multiscale texture synthesis algorithm that allows us to nse large neighbourhood systems to model some complex natural textures. As an added advantage of using the novel multiscale texture synthesis algorithm, phase discontinuities in the synthetic textures are reduced. Finally we show how the high dimensional representation of the texture may be modelled with lower dimensional statistics without compromising the integrity of the representation. The power Of oar modelling technique is evident in that only a small training image is required to derive respectable results even when the texture contains long range characteristics.