A Hybrid Fake Banknote Detection Model using OCR, Face Recognition and Hough Features

Currency duplication remains an emergent concern among nations due to the advancement of printing and scanning technology. Many note detection systems are present in banks but they are very costly and often inaccurate. The fields of image processing, neural network and machine vision have potential to significantly overcome this issue. This paper proposes a model comprised of Optical Character recognition (OCR), Face Recognition and Hough transformation algorithm. The microprinting, water-mark, and ultraviolet lines features of Bangladeshi notes are extracted for testing of genuine notes. The experimental results of the proposed model give accuracy as high as 93.33% which makes it suitable for deployment on a mobile application. Moreover, the obtained results are compared with the output from individual algorithm of OCR, Face Recognition and Hough transformation, to show that the proposed algorithm gives the highest accuracy.