This article presents a method to estimate the capture scale of a geographical database based on the characterization of its level of detail. This contribution fits in a larger research, dealing with the development of a general model to estimate the imprecision of length and area measurements computed from the geometry of objects of weakly informed geographical databases. In order to parameterize automatically a digitizing error simulation model, the characteristic capture scale is required. Thus, after a definition of the different notions of scales in geographical databases, the proposed method is presented. The goal of the method is to model the relation between the level of detail of a geographical database, by exploring inter-vertices distances, and its characteristic capture scale. To calibrate the model, a digitizing test experiment is provided, showing a clear relation between median intervertices distance and characteristic capture scale. The proposed knowledge extraction method proves to be the usefull in order to parameterize the measurement imprecision estimation model, and more generally to inform the database user when the capture scale is unknown. Nevertheless, further experiments need to be provided to improve the method, and model the relation between level of detail and capture scale with more efficiency.