Research On Geomagnetic Vehicle Detection And Classification Algorithm

At present, the automatic vehicle detection technology is the core technology of theoptimal control of road traffic and the implementation of highway toll system, which couldprovide information support to the application and development of intelligent transportationsystems. Therefore, the further research on the intelligent transportation is of greatsignificance. Research-based vehicle detection and identification of at home and abroad aswell as traffic optimization and control requirements, more in-depth research and explorationhave been done in detection equipments and detection methods. We adopt three-axisgeomagnetic sensor, combined with the characteristics of the vehicle geomagnetic detectionsignal, after wavelet filtering, calculation of vehicle parameters, extract models feature vector,cluster analysis. Then we judged car model by the fuzzy pattern recognition method. Wedesign a vehicle detection system based on three-axis geomagnetic sensor to provide real-timetraffic information and statistics, provide services for intelligent traffic optimal controlsystems, in order to optimize the management of road traffic.The vehicle detection system introduced by this paper could directly detect the stationaryvehicle, vehicle speed, vehicle length argument. To solve the problem of frequency drift ongeomagnetic detector, a base frequency drift suppression method based onsliding-time-window median filter clustering analysis has been presented, and set/resetmethod to eliminate the influence of the magnetic history. In this paper, we use the wavelettransform method for filtering the data to eliminate noise, geomagnetic curve closer to the realcurve. We should extraction feature after normalized treatment on the waveform curve, andthen make cluster analysis by kernel clustering methods, use normal distribution as the kernelfunction, calculate the corresponding parameters. Finally, we should transform the kernelfunction into membership function; classify the models with fuzzy rules. This method hasgood validity and feasibility of vehicle detection, vehicle identification rate.The main part of the paper describes vehicle detection technology status and the specificapplication and principles of AMR geomagnetic sensor in the detection of traffic flow. Weproposed the use of new sensors to detect vehicles, introduced wavelet filtering algorithmsand fuzzy pattern recognition method of vehicle, and analyzed the model identificationalgorithm, given the system’s hardware and software design.Finally, vehicle testing equipment was tested, also the data analysis and processing, afterrunning the software system, the overall performance is stable and reliable, able to achieve thedesired design effect. Then we summarized the advantages and disadvantages of this device, given solutions and the subsequent direction of development. Because of the use ofclient/server model, the system can easily increase equipment number or expand function, andis of great practical significance and wide application background for vehicle detection andvehicle identification.