INFORMYS - A FLEXIBLE INVOICE-LIKE FORM-READER SYSTEM

In this paper, we describe a flexible form-reader system capable of ex tracting textual information from accounting documents, like invoices and bills of service companies. In this kind of document, the extracti on of some information fields cannot take place without having detecte d the corresponding instruction fields, which are only constrained to range in given domains. We propose modeling the document's layout by m eans of attributed relational graphs, which turn out to be very effect ive for form registration, as well as for performing a focussed search for instruction fields. This search is carried out by means of a hybr id model, where proper algorithms, based on morphological operations a nd connected components, are integrated with connectionist models. Exp erimental results are given in order to assess the actual performance of the system.