Counterfeit Currency Detection
Counterfeit Currency Detection
Objective: To develop a system for recognizing fake currency that is not easily detectable by the naked eye. Traditional methods like chemical testing are impractical for widespread use, particularly among uneducated populations.
This system enables users to verify the authenticity of currency by simply taking a picture with a smartphone, providing an accessible and user-friendly solution to the increasing problem of counterfeit notes.
Extraction of Features
Extraction of features is the most important and challenging part. How efficiently we process the security features of the notes will itself determine the efficiency of our algorithms. Replicating these features is very difficult.
Method of Printing
Ink Properties
Thread
Watermark
Method of Printing: Intaglio printmaking technique is utilized for currency notes. It is very difficult to replicate the printed features because of special characteristics of the Intaglio print.
Ink Properties: The way a particular type of ink reacts to a particular kind of paper is unique and can be used to detect the genuinity or ingenuity of a note. The ink used in the intaglio printing technique will react to the currency paper differently as opposed to the ink used in counterfeit notes.
Thread: The registration of the currency notes and the text in the security strip are two important security thread-related features.
Watermark: The mean pixel intensity and standard deviation of the ROI are the two parameters considered for this feature.
Cinclusion
Methods Used:
Using MSE the average of the squares of all the errors and this is calculated between two images.The sample input notes were compared with these values and it was found out that the fake notes had a greater variance and mean square error.
SVM: The sample input notes were compared with these values and it was found out that the fake notes had a greater variance and mean square error.
Decision Tree: This process helps in filtering the currency notes belonging to a particular class.