DOCUMENT IMAGE-PROCESSING SYSTEM FOR NAME AND ADDRESS RECOGNITION

This article describes a real-time document image processing system. I ts objective is to recognize names and addresses from scanned address block images extracted from various tax forms of the United States Int ernal Revenue Service. The Name and Address Block Reader (NABR) system accepts both machine- and hand-printed address block images as input. Salient aspects of the system are presented, including document analy sis (connected component analysis, address block extraction, label det ection, hand-print/machine-print discrimination) and document recognit ion. Document recognition is performed in two nonidentical streams for machine-and hand-print; key steps are address parsing, character reco gnition, word recognition, and postal data base lookup (ZIP+4 and city -state-ZIP files). System output is a packet containing the results of recognition together with data base access status tile, Real-time thr oughput (8500 forms/h) is achieved by employing a loosely coupled mult iprocessing architecture where successive input images are distributed to available address recognition processors, The functional architect ure, software design, system architecture, and the hardware implementa tion are described. Performance evaluation on machine- and hand-writte n addresses are presented. (C) 1996 John Wiley & Sons, Inc.