This material is presented to ensure timely dissemination of scholarly and technical work. Copyright and all rights therein are retained by authors or by other copyright holders. All persons copying this information are expected to adhere to the terms and constraints invoked by each author's copyright. In most cases, these works may not be reposted without the explicit permission of the copyright holder.

Search for Publication

Year(s) from:  to 
Keywords (separated by spaces):

Bayesian Labeling of Remote Sensing Image Content

M. Schröder, K. Seidel and M. Datcu
The Eighteenth International Workshop on Maximum Entropy and Bayesian Methods, MaxEnt


In this paper we present a multi-level scheme for stochastic description of image content. The different levels are derived from the different degrees of abstraction. On the level of the image data, we use stochastic data models and Bayesian parameter estimation to derive low-level image features. On the next level, we derive meta features that provide both the fit of these models and the actual complexity of the data. The low-level and meta features are combined in an un-supervised clustering scheme to obtain an objective description of the image content. To obtain this objective description we use clustering by melting. The descriptions by several models are then linked to application-oriented, semantic labels using another process of Bayesian inference. We sketch in detail the various processes of inference and give an example for this kind of information on each level of abstraction using satellite images (RESURS-01 and X-SAR).

Download in pdf format
  author = {M. Schr\"oder and K. Seidel and M. Datcu},
  title = {Bayesian Labeling of Remote Sensing Image Content},
  booktitle = {The Eighteenth International Workshop on Maximum Entropy and Bayesian Methods, MaxEnt},
  year = {1998},
  pages = {1659--1669},
  editor = {Amin A. and Dori D. and Pudil P. and Freeman H.},
  publisher = {Kluwer},
  keywords = {remote sensing, image indexing, interpretation, model-based, knowledge based, texture, Bayesian statistics, Bayesian networks, clustering by melting, Gibbs fields, Markov random fields}