Yemeni Mobile Counterfeit Detection System Using Support Vector Machines, Fuzzy Logic and Image Processing Techniques

This paper proposes a novel easy to use real time mobile counterfeit detection system for the Yemeni currency. The need for such a system is necessitated by the vulnerability of the millions of common people to the ever advancing tricks of money counterfeiters. Some state-of-art algorithms and techniques were optimized and uniquely integrated into a single platform to develop the mobile phone counterfeit detection system. The system framework consists of four phases, the first of which requires acquisition of the currency image using a smart phone camera, the second involves pre-processing, the third involves feature extraction and the fourth involves the use of Euclidean distance to measure the similarity between the probe image and the image template from a database trained by support vector machines (SVM). The resultant Euclidean distances are then fed into fuzzy operator where the final similarity percentage is computed taking into account the priorities of the feature types. The feasibility of the system was tested experimentally using two Yemeni currency denominations and the results indicate a significant improvement in both currency recognition and counterfeit detection accuracy.