In this thesis, the model-driven software development paradigm is employed in order to support the development of Content-based Image Retrieval Systems (CBIRS) for different application domains. Modeling techniques, based on an adaptable conceptual framework model, are proposed for deriving the components of a concrete CBIRS. Transformation techniques are defined to automatically implement the derived application specific models in an object-relational database management system. A set of criteria assuring the quality of the transformation are derived from the theory for preserving information capacity applied in database design.