Nowadays, with the increase of resolution satellite images and the volume of geographic databases available, remote sensing images represent heterogeneous data sources increasingly needed and difficult to use and to exploit. These images are regarded as very rich sources and useful for updating cartographic databases and Geographic Information Systems (GIS). In order to keep up-to-date these databases, automatic change detection and data updating are required. Most of the existing works focus on the change detection between two images taken at different dates or between a digital map (such as vector or raster type) outcome of a database and an image. However, the detection of changes between multi-resolution satellite images and multi-representation spatio-temporal databases is seen as a problem of extracting features not only of low-level (segmentation, classification) but also high level (ontology, spatial reasoning, …). The aim of our study is to propose and realize a schema of system of analysis of remote sensing images for updating multi-representation spatio-temporal databases. These databases are GIS-like and Corine Land Cover (CLC) databases. This schema is based on methods of image processing (segmentation, classification, change detection, etc.), conceptual modeling of spatio-temporal databases, spatio-temporal ontology and spatial reasoning.