Reconhecimento de cédulas real usando algoritmo surf (speeded-up robust features)

Abstract

This paper proposes an algorithm to identify the Segunda Família do Real banknotes. The focus is to develop an easy and efficient algorithm that will be used for paper money recognition in order to help visually impaired people. The recognition system is a process that seeks to extract meaningful information from a particular image. Banknote recognition and detection technology intend to search for and extract the visible and hidden markings in the paper currency for a more efficient sorting. Each Real banknote (eg, 2, 5, 10, 20, 50 and 100 reais) has its own characteristics like color, size and texts. Often, money notes are blurred or damaged by constant use, which makes the identification difficult. Therefore, it is very important to select the right resources and the appropriate algorithm for this purpose. The method used in this work consists of a descriptor to provide greater speed and robustness in the extraction of points of interest on images. The method (SURF – Speed-Up Robust Features) is based on the Haar transform to reduce computational cost. The main advantage of this technique for extraction of characteristics is the identification of points of interest in an image at different scales and the possibility of matching points among them. The experimental results obtained prove the robustness, flexibility and efficiency of the proposed method.