Texture synthesis: a novel method for generating digital models with heterogeneous diversity of rock materials and its CGM verification

Huaiguang Xiao, Lei He, Xing Li, Qianbing Zhang, Wengui Li

Research output: Contribution to journalArticleResearchpeer-review

3 Citations (Scopus)


The interaction of meso-structures determines the macro-mechanical properties of rock material. Therefore, reliable description of heterogeneous diversity in rock is of great significance for studying its mechanical response and fracturing process. Based on the Markov random field theory, the texture synthesis method is optimized in terms of input exemplar and its best matching neighborhood size by using the mean shift clustering algorithm and the color histogram statistics. The optimized method can generate many parallel images of rock materials which have the function of feature replication and the characteristics of diverse heterogeneity. The reliability of the heterogeneous diversity models generated by the texture synthesis method is verified from three fundamental levels, including the CIELab color space, the geometric parameters of particles, and the numerical mechanical properties (CGM). The stress–strain curves and damage evolution processes of synthetic digital models under uniaxial compression have a strong characterization ability. Therefore, this new digital modeling method that represents the structural details and heterogeneous diversity of rock material can be utilized for evaluating the composition similarity and heterogeneous geometric features, and provide abundant materials for subsequent numerical mechanical analysis. This method provides a new idea for studying the meso-mechanical properties and failure mechanism of rock materials.

Original languageEnglish
Article number103895
Number of pages14
JournalComputers and Geotechnics
Publication statusPublished - Feb 2021


  • CIELab color space
  • Heterogeneous rock modeling
  • Markov random field
  • Numerical simulation
  • Texture synthesis

Cite this