Designing of a Currency Counting System with Integrated Counterfeit Currency Detection

Abstract

Counterfeit currency circulation remains a persistent threat to economic integrity, particularly in cash-dependent economies where manual verification is both time-consuming and error-prone. This paper presents the design and simulation of a currency counting system integrated with an automated counterfeit detection mechanism. The proposed system employs image processing techniques—including grayscale conversion, Gaussian blur, Canny edge detection, and Contrast Limited Adaptive Histogram Equalization—to extract discriminative features from scanned currency note images. Four feature categories are utilized: color histogram, texture, edge, and Oriented FAST and Rotated BRIEF keypoint features. A Random Forest classifier, trained on a labelled dataset of genuine and counterfeit Nigerian Naira note images, performs binary classification of each uploaded note. The system subsequently counts total notes, segregates genuine from counterfeit samples, and computes the aggregate monetary value of authenticated notes only. A web-based interface, developed using Streamlit, provides an accessible and interactive platform for real-time note scanning and result visualization. Experimental testing confirmed that the system correctly processes uploaded images, applies the trained classification model, and returns accurate counting and valuation outputs. The findings demonstrate that a software-based prototype integrating machine learning with image analysis can effectively simulate the core functions of a physical counterfeit-detecting banknote counter. Future work will incorporate ultraviolet, infrared, and magnetic sensor modules alongside hardware implementation using a microcontroller-driven mechanical platform.