Arabic Cheque Processing System: Issues and Future Trends

Abstract

From the administrative point of view, cheque processing involvesall tasks a bank officer may perform to process an incoming chequefor a client. This includes: accessing account numbers, verifyingnames and signatures on the cheque, verifying the date of the cheque,matching the legal amount with the courtesy amount and verifyingthe credit of the cheque writer. However, from the technical pointof view, cheque processing could involve capturing the cheque image,separating the foreground of the cheque from its background, extractingfields of interest and recognizing each of them. This work employstheories and methodologies from various fields ranging from Naturallanguage processing, Optical Character Recognition to Banking.The motivation of the work on Cheque processing is not less than themotivation of the entire research in artificial intelligence, whichaims to program the computer to carry out tedious routine processes,freeing time and space for humans to perform tasks that require higherlevels of intelligence. A major advantage of such study is that itcan be easily adjusted to serve more than 20 different countries(all of them use Arabic as their first language). In addition, legalamounts are widely found in documents other than bank cheques (e.g.business sell/purchase forms). Therefore, this study will be applicableto a wide range of applications. Moreover, similar languages (e.g.Urdu, Farisi) which use the same alphabet can benefit from thesestudies.The remaining sections provide a description of datasets availablefor researchers as well as a detailed description of one system thatprocesses legal amounts and one system dedicated for processing ofcourtesy amount