Researches On Several Key Technologies Of Banknote Detection Based On Multi-sensor Information Processing

With the increase of circulation and the promotion of globalization of Renminbi(RMB)dramatically,the level of automation and informatization of financial equipment such as currency counter,currency sorter,automated teller machine should be improved to make sure that the circulation and management level of RMB,combating the crime of counterfeiting banknote are effective;thus,the function of RMB and safety of financial system can be protected.This paper analyzes the shortcomings and defects of RMB banknote detection technology of Chinese financial industry at present.(1)Qualitative analysis and low level quantitative analysis are usually used for optical and magnetic feature identification.(2)Each sensor feature identification method is relatively isolated and the ability of information fusion is weak.(3)Intelligent algorithm with excellent performance is highly complex and poor real-time.Aiming at the second defect,this paper focuses on several key technologies of banknote detection based on multi-sensor information.The main content of this paper contains the construction of verification platform for multi-sensor banknote detection and the study of banknote image processing algorithm based on multi-sensor information.The work of the construction of verification platform for multi-sensor banknote detection contains the construction of electro-mechanical platform,software and hardware design of detection and control system.Based on the verification platform,it realizes that the transmission mechanism of platform transfers banknote and makes it to pass through each sensor module under the control of the detection and control system.It realizes dynamic acquisition of the original data of the security feature information of banknote,such as optical,magnetic and image features,and transmits the sampled data to the PC and save it.It provides data for the study of off-line simulation algorithm for analysis and identification of banknote security features.The detection and control system includes electro-mechanical control,optical-magnetic features detection system and banknote image acquisition system based on FPGA and CIS.Subsystems make full use of their control interface and communication interface to achieve collaborative work and multi-sensor identification information sharing.It supports for physical simulation research of multi-sensor information fusion algorithm.In the study of banknote image processing algorithm based on multi-sensor information,it detects the tilt angle of banknote image based on the weighted least square method and corrects banknote image by shear transformation algorithm,which ensures that the banknote image feature area,such as serial number,can be segmented correctly.Compared with the random Hough transform and normal least square method,the results shows that weighted least square method has advantage of less computation cost and high stability,which is able to correct the tilt banknote image fast.Based on the banknote image tilt correction and in the condition of the multi-sensor identification information sharing in the verification platform,it researches on fast segmentation of serial number image and character recognition algorithm of banknote image based on banknote denomination and face direction information from magnetic feature identification.A simple algorithm of banknote feature area segmentation based on multi-sensor information can be port to banknote image acquisition system which is based on FPGA.It can reduce the processing task of the DSP processing unit in typical embedded banknote image processing system.This paper focuses on the construction of verification platform for multi-sensor banknote detection and the study of banknote image processing algorithm based on multi-sensor information to study of banknote detection technology based on multi-sensor information.It provides the original data and physical simulation platform for the study of multi-sensor information fusion algorithm.It attempts the preliminary study of multi-sensor information fusion algorithm to simplify banknote image processing algorithm and process,to balanced assign image processing task of each processing unit in embedded banknotes image processing system,to improve the ability of collaborative work and the efficiency of banknote image processing of the system.