Dynamic texture segmentation

Author

Doretto, Gianfranco ; Cremers, Daniel ; Favaro, Paolo ; Soatto, Stefano

Author_Institution

Dept. of Comput. Sci., UCLA, Los Angeles, CA, USA

fYear

2003

fDate

13-16 Oct. 2003

Firstpage

1236

Abstract

We address the problem of segmenting a sequence of images of natural scenes into disjoint regions that are characterized by constant spatio-temporal statistics. We model the spatio-temporal dynamics in each region by Gauss-Markov models, and infer the model parameters as well as the boundary of the regions in a variational optimization framework. Numerical results demonstrate that - in contrast to purely texture-based segmentation schemes - our method is effective in segmenting regions that differ in their dynamics even when spatial statistics are identical.

Keywords

Gaussian processes; Markov processes; computer vision; image segmentation; image texture; natural scenes; optimisation; variational techniques; Gauss-Markov models; disjoint regions; dynamic texture segmentation; image sequence; model parameters; natural scenes; numerical results; spatial statistics; spatiotemporal dynamics; spatiotemporal statistics; variational optimization framework; Computer science; Gaussian processes; Image segmentation; Layout; Marine vehicles; Mobile robots; Remotely operated vehicles; Statistical distributions; Statistics; Vehicle dynamics;

fLanguage

English

Publisher

ieee

Conference_Titel

Computer Vision, 2003. Proceedings. Ninth IEEE International Conference on

Conference_Location

Nice, France

Print_ISBN

0-7695-1950-4

Type

conf

DOI

10.1109/ICCV.2003.1238632

Filename

1238632