Nearly all three-dimensional reconstruction methods lack proper model knowl
edge that reflects the scene. Model knowledge is required in order to reduc
e ambiguities which occur during the reconstruction process. It must compri
se the scene and is therefore complex, and additionally difficult to acquir
e. In this paper we present an approach for the learning of complex model k
nowledge. A (large) sample set of three-dimensionally acquired buildings re
presented as graphs is generalized by the use of background knowledge. The
background knowledge entails domain-specific knowledge and is utilized for
the search guidance during the generalization process of EXRES. The general
ization result is a distribution of relevant patterns which reduces ambigui
ties occurring in 3D object reconstruction (here: buildings). Three differe
nt applications for the 3D reconstruction of buildings from aerial images a
re executed whereas binary relations of so-called building atoms, namely te
rtiary nodes and faces, and building models are learned. These applications
are evaluated based on (a) the estimated empirical generalization error an
d (b) the use of information coding theory and statistics by comparing the
learned knowledge with non-available a priori knowledge.