On the Possibilities of Using Traditional Pattern Recognition Techniques in Brazilian Banknote Characterization

In Brazil there are seven kinds of banknotes where each one has a specific color associated to its correspondent value. Analysing the Brazilian banknote design, one can see that there are many different regions. In the region named Puzzle, which is formed by simple patterns present in both sides of banknotes, a simplified process of pattern recognition and classification can be proposed. That is, as each note presents a diverse pattern and color, such elements can be investigated to be used for recognizing their values. This paper aims to help with the accessibility for handling money in everyday life, studying the possibilities of classic image analysis to develop an application to improve the social inclusion of visually impaired people. To this end it is proposed to combine the use of Template Matching Technique and HSV Space Color. The Normalized Correlation Coefficient (NCC) was verified to be the best method, after the evaluation of six other methods using the OpenCV library tools and Python computer language. According to achieved results, it is possible to stipulate the value of banknotes associating the characteristic of hues and NCC using the here presented and justified approach.