Abstract:
A method for counterfeit bill detection based on a digital image for mass-production smartphones is developed. The method under consideration does not require new protective elements to be designed and introduced into the print and is based on the use of digital image analysis and recognition methods, allowing one to carry out an automatic search and verification of known protective elements of the print. The peculiarity of the proposed approach is associated with constructing a feature vector for each type of samples and their subsequent classification using machine learning based on a training sample. The method is realized as a program application for smartphones, performing the automatic detection of an object in the frame, shooting the cameracaptured object, rejection of unsuitable images, determination of a face-value and type of the banknote, and finally, verification of the authenticity.