An assistant to the modelisation of forms

This article deals with a form description module based on structure extraction. This description module helps a user for the creation of models used in a document treatment system. The aim of these model creation module is to define the filled up areas of forms used for the interpretation of the form content. This document treatment system will be applied on forms received by the CAF-Caisse d´Allocations Familiales-the French national family allowance Department. The form description module is composed by 3 main stages, from acquisition to structuration. The first one corresponds to the acquisition and the correction of the potential skew detected on the scanned image. The second extracts the informative features contained in forms. The last one organises the different features to obtain a hierarchical structuration of forms