A Discriminating Banknotes Counterfeit Research Based On Multi-spectral Image Analysis

Banknote multispectral image security identification technology has been the hot spot in the field of paper currency anti-counterfeiting,relating to the national financial security.The research has great theoretical research value and widespread application background.Banknote image contains the image information of visible light,infrared light and UV light.Through analysis and understanding of the notes multispectral image,we can classify these banknotes high reliably.Also we can stably detect the fake money that the traditional anti-counterfeit technology is unable to distinguish,as well as the money that are altered and in small circulation.Thus guarantee the security and reliability of the currency notes.In this paper,we make the in-depth study in the banknote image acquisition,image preprocessing,feature extraction,classification algorithms and other key technologies.Summarized in the following aspects:(1)In this paper,we solve the acquisition technology of the Multispectral banknote image.This thesis introduced the mainstream imaging sensor,compared their advantages and disadvantages.For the advantages of low cost,wide market,convenient installation and so on,CIS sensor is selected.In addition,this thesis introduced the acquisition module design principle and the working process.(2)For the Multispectral banknote image,an improved Gray-computing algorithm is presented to satisfy the system inspection speed.Preprocessed the illumination compensation,denoising,edge detection,image rotation,finally,a Standard image is got.(3)As a global feature extraction method,PCA feature will not reflect the nonlinear manifold of high-dimensional data.when the light changed,the recognition rate will be significantly reduced.This thesis introduced an adaptive LPP algorithm to extract the key features to improve the accuracy.This algorithm,not only,to solve the traditional linear methods like PCA and others,which have shortcomings of keeping the original data been the non-linear manifolds,but also solve the disadvantage of the non-linear method to obtain a new sample point disadvantage low-dimensional projection.(4)BP neural network algorithm has become a mainstream classification algorithm.It is widely used in various industries,and has its unique advantages.In this paper the banknote image characteristics were processed in two steps,firstly,training the sample image,and then,test the rest of the samples.At last,experimental results show that the methods are effective.