Philippine Currency Counterfeit Detector using Image Processing

Abstract

Image processing is the utilization of a computer to process digital images through an algorithm. This method can also be used in detecting the security features of banknotes to tell if the banknote is authentic or counterfeit. Using algorithms such as Canny Edge Detection, Hough Line Transform, Optical Character Recognition and K-Means Clustering to detect Philippine currency banknote level 1 security features such as watermark, asymmetrical serial number, see-through print, and security thread. Canny Edge Detection would be used to detect edges and curves found on the banknotes. Hough Line Transform would be used to detect straight lines found on the banknotes. Optical Character Recognition (OCR) would be used to distinguish text found on the banknotes. K-means clustering would be used to detect color ranges used by the banknotes by means of vector quantization. The tests yield a result of 100% for 200-peso banknotes, 97.50% for 500-peso and 1000-peso banknotes, 96.67% for 50-peso banknotes 91.82% for 20-peso banknotes, and 91.67% for 100-peso banknotes, which has a mean average of 95.86% which is significantly higher than the average accuracy from previous studies of 86.27%.