Self-service automatic teller machines (ATMs) have dramatically altered the ways in which customers interact with banks. ATMs provide the convenience of completing some banking transactions remotely and at any time. AT&T Global Information Solutions (GIS) is the world's leading provider of ATMs. These machines support such familiar services as cash withdrawals and balance inquiries. Further technological development has extended the utility and convenience of ATMs produced by GIS by facilitating check cashing and depositing, as well as direct bill payment, using an on-line system. These enhanced services, discussed in this paper, are made possible primarily through sophisticated optical character recognition (OCR) technology. Developed by an AT&T team that included GIS, AT&T Bell Laboratories Quality, Engineering, Software, and Technologies (QUEST), and AT&T Bell Laboratories Research, OCR technology was crucial to the development of these advanced ATMs.

A neural network algorithm-based system that reads handwritten ZIP codes appearing on real US mail is described. The system uses a recognition-based segmenter, that is a hybrid of connected-components analysis (CCA), vertical cuts, and a neural network recognizer. Connected components that are single digits are handled by CCA. CCs that are combined or dissected digits are handled by the vertical-cut segmenter. The four main stages of processing are preprocessing, in which noise is removed and the digits are deslanted, CCA segmentation and recognition, vertical-cut-point estimation and segmentation, and directly lookup. The system was trained and tested on approximately 10000 images, five- and nine-digit ZIP code fields taken from real mail.< >