* Corresponding author OBJECT-BASED CHANGE DETECTION AND CLASSIFICATION IMPROVEMENT OF TIME SERIES ANALYSIS Dirk Tiede a* , Annett Wania b & Petra Füreder a a University of Salzburg, Centre for Geoinformatics - Z_GIS, Salzburg, Austria - dirk.tiede@sbg.ac.at; petra.fuereder@sbg.ac.at b Annett Wania, European Commission Joint Research Centre, Institute for the Protection & Security of the Citizen (IPSC), Global Security and Crisis Management Unit, Ispra (VA), Italy - annett.wania@jrc.ec.europa.eu KEY WORDS: post-classification change detection, OBIA, CNL, object comparison ABSTRACT: The paper presents a new approach for post-classification change detection. Classification results are integrated in an object-based hierarchical knowledge framework, compared and aggregated on a change detection layer. The approach is - dependent on the complexity of the input classifications - semi-automated and transferable. The change detection framework is illustrated by two different applications. The first application is a change detection analysis of land use / land cover classifications which aims at identifying changes in land use since the land reform in Zimbabwe. The second application analyses changes in built-up area to reflect the urban development of the city of Harare, Zimbabwe. 1. INTRODUCTION Many change detection approaches are documented in the literature (see for example Coppin et al., 2004), also in the area of post-classification methods. Usually pixel-by-pixel or object- by-object comparison is used in post-classification change detection, resulting in a complete (and complex) matrix of change. In this paper we introduce a new approach using a topologically enabled object-by-object comparison, where changes are aggregated to a change detection layer. The resulting layer is an easy to use quantification and visualization of relevant changes. The approach is based on object-based image analysis (OBIA), a well-established methodological framework for integrated image analysis (Blaschke, 2010). The research was conducted within the frame of the FP7 GMES project G-MOSAIC (GMES Services for Management of Operations, Situation Awareness and Intelligence for Regional Crises) and is illustrated by means of two application examples. Figure 1. Object-based post-classification change detection framework 2. METHODOLOGY Existing multi-temporal (e.g. land cover) classification results (vector or raster data) are integrated in a hierarchical knowledge framework, which is flexible in terms of (1) number of classes, (2) aggregation of classes and (3) number of time slots to be analysed. The different classification results are embedded in a topological enabled hierarchy (topological relationships between the objects in vertical and horizontal direction), in which they are compared and aggregated to a change detection layer (e.g. a user defined regular gridded layer). The knowledge Proceedings of the 4th GEOBIA, May 7-9, 2012 - Rio de Janeiro - Brazil. p.223 223