Issue No. 08 - August (1998 vol. 20)
DOI Bookmark: http://doi.ieeecomputersociety.org/10.1109/34.709593
<p>Abstract—We present a novel optimization framework for unsupervised texture segmentation that relies on statistical tests as a measure of homogeneity. Texture segmentation is formulated as a <it>data clustering</it> problem based on sparse <it>proximity data</it>. Dissimilarities of pairs of textured regions are computed from a multiscale Gabor filter image representation. We discuss and compare a class of clustering objective functions which is systematically derived from invariance principles. As a general optimization framework, we propose deterministic annealing based on a mean-field approximation. The canonical way to derive clustering algorithms within this framework as well as an efficient implementation of mean-field annealing and the closely related Gibbs sampler are presented. We apply both annealing variants to Brodatz-like microtexture mixtures and real-word images.</p>
Image segmentation, pairwise clustering, deterministic annealing, EM algorithm, Gabor filters.
Thomas Hofmann, Jan Puzicha, Joachim M. Buhmann, "Unsupervised Texture Segmentation in a Deterministic Annealing Framework", IEEE Transactions on Pattern Analysis & Machine Intelligence, vol. 20, no. , pp. 803-818, August 1998, doi:10.1109/34.709593