Research On The Identification System Of Bank Notes

Although the electronic bills are used for a long time, it can not meet completely the need of market entrants under the background of existing market and supervising.Therefore, the existence of papery bills are very necessary. With the development rapidly of society, the business of bank increase drastically everyday.If the bills are classified, inputed, checked by people, the result is that the vast manpower and material resources will be wasted. In addition, the process is easy to be wrong. The system of bills automatic identification which meet to the requirement of the electronization of banks is designed and researched. As a result, the flow becomes simplified, the efficiency of work is improved, and the risk of finance is prevented.The system of bills automatic identification consists of handling the image of the bills beforehand, locating discriminating unit, and distinguishing and export the element in the bills which contain serial number, amount of money, signature and seal of accessory. The main work of the thesis is as follow:(1) Bill pretreatment. The process includes binarization, skew correction and de-noising. The accuracy of futher fix and recognition is improved through adding the filtering link.(2) Bill classification. The bank bills are various, such as the evidence of deposit, the evidence of business and check online, and so on. Therefore the bills are classified through matching the model based on the image size, the characteristics of layout, the form line, the statement heading and the text area. The Harr characteristics and the algorithm of Adaboost is applied to test the text area.(3) Recognition unit location. The method of moving datum point which is located in the link of locating discriminating unit is proposed. The problem of holistic excursion is solved effectively when bills are printed outright.(4) The characters are partitioned by the method of projection. The characters are recognized by the BP neural network after training. The seal of accessory is recognized based on the module of colour RGB.The bank’s bills are recognized automatically in the programming environment of VC++2010. The results show that therecognition rate is about 91%, and the speed of recognition less than 1 second. The requirement of recognizing automatically of bank’s bills achieves the standard.