Conference paper

Scarpa G., Haindl M.

Proceedings of the 18th International Conference on Pattern Recognition. ICPR 2006

IEEE Press

Hong Kong

2006

151-154

0-7695-2521-0

1051-4651

prof. Ing. Michal Haindl, DrSc. (ÚTIA - Oddělení rozpoznávání obrazů)

texture segmentation, unsupervised classification

A novel color texture unsupervised segmentation algorithm
is presented which processes independently the spectral
and spatial information. The algorithm is composed
of two parts. The former provides an over-segmentation of
the image, such that basic components for each of the textures
which are present are extracted. The latter is a region
growing algorithm which reduces drastically the number
of regions, and provides a region-hierarchical texture
clustering. The over-segmentation is achieved by means of
a color-based clustering (CBC) followed by a spatial-based
clustering (SBC). The SBC, as well as the subsequent growing
algorithm, make use of a characterization of the regions
based on shape and context. Experimental results are very
promising in case of textures which are quite regular.