Estudo da detecção de cédulas falsas da segunda família do Real Brasileiro através de técnicas de processamento de imagens

Abstract

With the advent and creation of a oficial currency for the national territory, attempts do defraud Brazilian paper money and forge new ways to counterfeit it also emerged, which only grow and evolve over the years. In this way, considering the utmost importance of money for the functioning of the economy and society as a whole, it i salso necessary to validadte it properly and combat attempts at counterfeiting, both by human operators and automatically. This paper proposes a system capable of automatically separating and classifying Brazilian Real bills from the Second Family into a group of counterfeit bills and another group of genuine bills. For this, five security features of Brazilian paper money are selected, namely the “Watermark”, “Color-Changing Number”, “Holographic Strip”, “Hidden Number”, and “Fluorescent Elements”, studied individually using different image acquisition and processing techniques, image mathematical analysis, and the determination of a classification criterion. The methodology chosen for this study uses MATLAB and takes into account methods used in other works published from the perspective of Indian currency. This experiment presented na accuracy of 100% for four of the five security features tested, and 75% for the last, which led to the discussion of future suggestions for development and improvement for the Project, also exposed in this paper.