Paper Currency Distribution System Research Based On The Theory Of Contourlet Transform

Banknote is the symbol of a country, paper money identification and classificationis an indispensable part in a variety of fields, such as the bank deposit and withdrawalmachines, rail transportation automatic ticket vending machines, enterprises vendingmachines, counterfeit detection machines which are used to combat crime by thepublic security organs. With the rapid development of Chinese economy, the financialcrime is rampant, so the paper money recognition and anti-counterfeiting technologyhas become a focus of research scholars at home and abroad. Paper note image notonly contains the information of multi-spectrum, thickness and magnetic signal, butalso contains a lot of texture information. At the same time, the paper money itself hasa serial number which can identify each note and provide a favorable basis for notescounterfeit detection. This paper mainly analyzes the note image feature extractionbased on Contourlet transform domain and the related technology of paper serialnumber identification. And it also builds a note image classification and identificationsystem prototype. This method proposed in this paper can overcome the problem of thelow quality, low defect rate of the banknote image recognition. Concrete researchcontent is as follows：1、This paper proposes a new note image feature extraction method based onContourlet transform. First of all，The Contourlet transform is applied to the originalbanknote image after preprocessing, And the paper currency image is decomposed intomulti-scale and multi-direction images which then become the high and low frequencysubbands. Then some of the most representative statistical properties, which canrepresent the distribution of coefficients, are extracted, and put these statistics togetherwhich will constitute notes image feature vector. Finally, the BP neural networkclassifier be used for paper currency image recognition. This method have highstability and reliability and can effectively overcome the shortcomings of the lowrecognition rate of low quality paper currency image.2、This paper proposes a banknotes image sequence number recognition methodbased on Contourlet transform. First in obtaining character stage, it will Complete thepreprocessing by combining the prior knowledge, projection and scanning methodsand implement effective positioning and segmentation of the serial number. In the stage of character feature extraction, for the first time the Contourlet theory isapplied to the paper serial number in the field of feature extraction in this paper. Finally,for paper serial number is made up of two parts of the letters and numbers, to solve theproblem of numbers and some characters are hard to distinguish, In this paper, arational logic method is used to solve the problem of numbers and some characterswhich are hard to distinguish,and ultimately achieve the purpose of effectively identifythe serial number after using BP network for character recognition process.3、In this paper, banknote processing systems frame was designed, the softwareand hardware structure of the system is analyzed, and its working principle andalgorithm was analysised specifically. It builds a development platform for the imageprocessing method based on frequency domain, which proposed in this paper.