Multilevel Analysis to Recognize Original Voucher from Faked Voucher

Abstract

Voucher documents have become a very important information carrier in daily lives to be used in many applications. A certain class of people could exploit the trust and indulge in forging or tampering for short or long term benefits unlawfully. This holds a serious threat to the economics and the system of a nation. The aim of this paper is to recognize original voucher document through its contents. Forgery of voucher document could have serious repercussions including financial losses, so the signature, logo and stamp that are used to determine being a genuine or not by using multilevel texture analysis. The proposed method consists of several operations. First, detection and extraction of signature, logo and stamp images from original voucher document by using auto crop method. Second, each image is processed in allotted level. Third, the voucher document is classified depending on a result of each level to determine being a genuine or not. Accuracy of 94% for identification process and 95% for verification process were achieved.