The Design Of AM-EAS System And The Research Of Label Detection Algorithm

The electronic article surveillance system(called EAS)is a kind of high-tech anti-theft devices in the retail business,it can improve the sales profit by reducing the stolen goods.Due to the development of science and technology,the working environment of EAS system is gradually more and more complex,the detection performance of the EAS system is seriously affected with all kinds of electromagnetic interference.The acoustic magnetic(called AM)technology provided a new path in EAS system,AM-EAS system was designed according to the principle of tuning fork,it’s performance better than other types of products.The paper would design a AM-EAS system based on DSP,and study a detection method of AM-label with support vector machine(called SVM)based on the AM-signal characteristics.In the paper,the detector of AM-EAS system was designed based on the TMS320F28335 as the core,working frequency of the system is 58 kHz.By means of introducing a sync signal(NET),the system that could control the doorways of multiple antennas was designed by the method of time division.The paper expounds the basic composition and the operating principle of AM-EAS,and it analyzes mathematical model of the acoustic magnetic electronic tag;.The detector is introduced about the system hardware principle diagram and the function of each circuit module in detail,including the model of signal driver,the model of signal receiver,the model of communication.The paper summarized and corrected the problem of PCB layout drawing in the end.In view of the kind of situation that the traditional label recognition algorithm would make the system detection rate and anti-interference problem poorer in AM-EAS,In this paper,a new signal detection algorithm that was the combination of a kind of the improved artificial fish swarm algorithm(called IAFSA)and SVM had been proposed,the principle of SVM and the traditional AFSA were introduced in detail.It pointed out some problems that exist in SVM,some improved suggestions were put forward according to the shortage of the AFSA.The experiment collected enough label sample data,through the way of digital signal processing,the paperextracts and analyzes the characteristics of the system sound and the AM tags signal,after the comprehensive consideration,it selects the amplitude,frequency and signal-to-noise ratio about the signal as the input vector of SVM.Finally,according to the scheme of system,AM-EAS was designed with signal transceiver one-piece and the label recognition algorithm,the pap-er tested the performance of the system hardware and algorithm in the end,the system hardware module works well as same as the preset parameters.The tag identification algorithm based on SVM detection distance of 1.5 m,the detection rate was 99.75%,average algorithm caused 4.6 ms,it could meet the requirements of real-time detection system.