Change detection is a major application domain for image analysis techniques in remote sensing. Besides the analysis of multitemporal
imagery there is also the need to update or revise previously created thematic data with the help of recently acquired
imagery. The emergence of object-based image analysis provides another approach to the change detection problem and derived
methods and algorithms have been applied to it accordingly. Simultaneously there is continuous demand for techniques that produce
results that are GIS-ready. In this study we put forward a proposal for automated change analysis for thematic data using an objectbased
image analysis methodology. It is implemented within the scope of IMALYS, a software system able to conduct a
comprehensive object-based investigation of remote sensing imagery including segmentation, feature retrieval and classification.
Segmentation is conducted based on a hybrid approach with the emphasis on a low-degree-of-freedom system. Retrievable attributes
for the created segments include a number of built-in features as well as ones defined by the user. For identification of real-world
objects the presented software implements a two-stage process consisting of an unsupervised classification based on self-organizing
maps and a supervised classification based on a context-sensitive object model. For change detection the latter process can be
provided with training data based on existing thematic mappings. Using statistical analysis the presented software is capable to relate
the underlying imagery to previous mappings or classifications results. The result of this process is a measure of probability of a
certain segment to be still part of the previously assigned class. As this process is based on statistical calculations it is not only
suitable to be applied on large areas and data volumes but also dependent on such conditions. The introduced software is integrated
in a GIS-environment and its results can be therefore easily incorporated in following workflows and processes.