This article deals with the description of a module for information structure extraction. This module is applied to forms which are used by the CAF, the French national family allowance Department -Caisse d'Allocations Familiales-. The aim of this module is to create a base of information structures - models - defining their forms and their content so as to be able to treat them automatically. The module deals with the various stages of form treatment, from acquisition to modelisation. It is composed by 3 different stages. The first corresponds to low-level processing -i.e., binarisation, skew correction-. The second extracts the informative features contained in the forms. The last one organises the different features to obtain form modelisation thanks to a hierarchical structuration. The creation of this base of information structures will be used for a second module for type form identification based on the comparison of these information structures.