Design And Implementation Of The Notes Image Acquisition System Based On Multi-core ARM Processor

Counterfeit currency is an international problem,and banknote identification is the basic function of most financial instruments.Traditional financial instruments use magnetic signal analysis,infrared feature analysis,fluorescence response analysis and other methods to realize the identification of banknotes.With the development of technology,the notes based on multispectral image analysis become the necessary function of high-end financial instruments,so that the notes image acquisition system is an important module of it.At present,cash-counter and cash-sorting machine in the market are mostly based on DSP and FPGA design,it has good performance,meet the needs of real-time multispectral image acquisition and analysis,but the comprehensive cost of this scheme is higher,limiting the image analysis function in the field of financial equipment popularization.In maintaining the existing function and performance,reducing the cost,getting the multispectral image analysis function to the low-end civilian implements,is a important topic of current financial instruments industry.Based on multi-core ARM processor,this paper designs and develops a new low-cost paper money image acquisition system to solve the problem of high cost of financial equipment.Firstly,based on ARM processor,CPLD programmable logic device and CIS image sensor,the paper designs the overall technical scheme of the notes image acquisition system.Based on image sensor,analog-to-digital converter and CPLD timing logic,the line array image acquisition module is realized,the ARM processor collects data through high speed video interface and carries on the real-time notes multispectral image processing and analysis.Secondly,based on the technical scheme,the hardware circuit design of the notes acquisition system was completed,the impedance matching,the power filter or isolation were emphatically solved in the circuit design.Then,using VHDL hardware programming language,the timing design of CPLD is accomplished,including image sensor,A/D converter,ARM video Interface timing control,and related communication logic;Finally,based on the embedded Linux platform,the driver and the state machine design are accomplished in the driving layer,and the efficient image preprocessing scheme and message queue structure design are accomplished in the application layer.Finally,in the real environment of the cash-counter machine,this paper test the notes image acquisition system,the results show that the system works stably and the image quality is good.At the 100 DPI resolution,the acquisition of the notes image in the infrared,white of front and back,totaly 4 pairs of images,real-time image acquisition speed up to 900/min,meet standard of A-class cash-counter machine.At the same time,the computational performance of the ARM processor can satisfy the demand of real-time image recognition and banknotes number identification.