The Vehicle Identification Technology Based On BP And HOG Apply To The Wireless Devices

City transportation monitoring system is an important branch of intelligent transportationsystems,as a core function of the intelligent development in the city transportation,it occupiedan irreplaceable position in the transport sector and pattern recognition directions. The motorvehicles’ number is increasing with the improvement of people’s living standards.Especiallythe private vehicles’ ownership is rapid growth,it is bound to trigger a series of trafficproblems in this situation.The growing number of traffic accidents and transportation violateregulations also gave us severe tests.These problems fully exposed the manage issues andholes from the transportation agencies,as the reality of the situation can not be achieve humanmonitor in the24hours on all the sections and junctions of the cities,to solve this problem wemust using the technology power.Using the method based on machine vision and imageprocessing technology combination,to detect and identify the license paltes,logo and modelson a car accurately and correctly through building an intelligent transportation monitoringsystem.Currently there are many relatively mature vehicle identification technology,in each ofthe charging system in the major developed countries.Parking payment is no longer such aclumsy and cumbersome way,by using the vehicle device and the ground station equipmentmutual recognition approach that improved the payment efficiency of the toll stations.Anotherrelatively common way is to be detected by the intrinsic parameters of the vehicle, italways applying an appropriate classification algorithm in certain vehicle classificationstandards with the aid of microwave, laser, red, SAW and other ways to identify the vehicleinformation. We call the method is automatic vehicle classification,due to this approach isrelatively easy,so it is widely used in the daily life.Many approach derived in this way such asgeomagnetism detection, microwave detection, laser infrared detection, ultrasonic detection,loop detector, etc.While this approach have a relatively high recognition accuracy, but due toits installation process is relative complexity,and sometimes it also easily be damaged andhuge expenses by environmental influences,even a traffic gridlock.In this paper, we employ a more advanced pattern recognition and image technology, It’sa more mature and advanced technology in the field of graphic images to identify the vehicle’s logo and the vehicle’s model.Taking into account the actual needs of living,we splitthe specific vehicle identification into two bodies, one is to locate and identify the vehicle’slogo and classificate accurately in accordance with the license plate information.Another oneis to identify the vehicle, while the vehicle manufacturer has been determined, next is toidentify a specific model of the vehicle in the manufacturer.Finally, put the system onto theAndroid platform according to the characteristics of the vehicle, the method can be appliedcorrectly on mobile devices by the captured image recognition and classification on theAndroid platform, which has a positive meaning in the intelligent vehicle identification.In the identification process of a vehicle’s logo,firstly we studied the method of vehiclelocation method, we employ the elimination method to detect the background image of avehicle, according to the position relationship between the logo and the license plate of the carmarked a rough positioning, and then employ edge strength and morphological analysis todetermine the precise target area of the car.Then depending on the prior knowledge of thetexture subject of car’s brand,we classify the logo into the BP neural network by extractingSIFT feature.Finally, the two BP neural networks: prediction network and verify network thatis completed vehicle identification.In this paper we select a lot of cars’ logo to identify,results show that the algorithm has higher recognition rate and robustness to light and noise.In order to overcome difficulties which come from models of the same brand car have a littlechanges in the identification phase.This paper draws on the recent application of gradientorientation histogram in target detection.Meanwhile, in order to improve the overall efficiencyof the vehicle recognition,we proposed to use of support vector machine theory to identify acar.It will have invariance in the respect of translation, rotation and scale based on the featurehistogram,it will be used to determine the optimal SVM classification of surface and therebyidentify the cars model.Experimental results demonstrated that choose SVM classifier notonly improve the recognition efficiency, while reducing the classification time.Finally, the identification technology of vehicle used on Android intelligent mobileterminals,through testing and identification the pictures from actual life,our vehicle system tomeet the needs of users,we can identify the vehicle’s logo and model under either weather orcomplex environmental backgrounds accuracily.