An Optimization Approach to Unsupervised Hierarchical Texture Segmentation

In this paper we introduce a novel optimization framework for hierarchical data clustering and apply it to the problem of unsupervised texture segmentation. The proposed objective function assesses the quality of an image partitioning simultaneously at di erent resolution levels and yields a sequence of consistently nested image segmentations. A novel model selection criterion to select signi cant image structures from various scales is proposed. As an e cient deterministic optimization heuristic a mean eld annealing algorithm is derived.

• Type: Conference
• Year: 1997
• Where: ICIP
• Authors: Thomas Hofmann, Jan Puzicha, Joachim M. Buhmann
• Keywords: Cient Deterministic Optimization, ICIP 1997, Image Processing, Mean Eld Annealing, Unsupervised Texture Segmentation