Fuzzy Logic Weighted Averaging Algorithm for Malaysian Banknotes Reader Featuring Counterfeit Detection

Abstract

This paper proposed a novel fuzzy logic weighted averaging (FLWA) algorithm in image processing techniques to detect counterfeit Malaysian banknotes. Image acquisition techniques on banknote position detection and re-adjustment, image pre-processing techniques, feature extraction methods on Malaysian banknotes’ watermarks are also covered in the paper. The FLWA Algorithm has the advantage of a much simpler model since it is a human guidance learning algorithm that does not require enrolment process to get the specific weights for each security feature. Each security feature is treated with equal weight. The experimental results also shown that FLWA model also outperform the MobileNet model and VGG16 model in Malaysian banknotes’ counterfeit detection. It has a distinct advantage over earlier or current banknote counterfeit detection techniques in that it adopted the known watermarks features, with known machine learning techniques to identify real Malaysian banknotes and detect those counterfeit Malaysian banknotes.

Keywords

• image processing
• fuzzy logic
• malaysian banknotes
• banknote reader
• counterfeit detection