Department of Pattern Recognition

(eng): A novel generative colour texture model based on multivariate Bernoulli mixtures is proposed. A measured multispectral texture is spectrally factorised and multivariate Bernoulli mixtures are further learned from single bit planes of the orthogonal monospectral components and used to synthesise and enlarge these monospectral binary factor components. Texture synthesis is based on easy computation of arbitrary conditional distributions from the model. Finally single synthesised monospectral texture bit planes are transformed into the required synthetic multispectral texture. This model can easily serve not only for texture enlargement but also for segmentation, restoration, and retrieval or to model single factors in complex Bidirectional Texture Function (BTF) space models. The strengths and weaknesses of the presented Bernoulli mixture based approach are demonstrated on several colour texture examples.