Based On Models Of Sequence Image Recognition Technology Research

Intelligent Transportation system is the development trend of traffic for the current century,vehicle recognition is an important part of ITS,and also is one of the key technology.As the traditional method has the disadvantage of high cost and trouble in construction,the way of vehicle recognition bases on video image can overcome the above short comings, so widely use in various aspects.This paper studies the technology of vehicle recognition which bases on sequence image, using statistical background updating method and three frame difference way base on speed estimation to vehicle detection,the way of local statistic and project technology for vehicle location.It makes the way of KNN and MDC to classify for sequence image which according to the geometrical characteristics of vehicle and the algebra features,and displaying the result in MFC,the main content as follows:(1)Vehicle detection. After image preprocessing,this paper make use of statistical method to updata the background,it gets better results for detection;At the same time,it carries out a method that a kind of the speed of vehicle make use of three frame difference method which overcome image empty phenomenon because of continue frame in traditional method.(2) Vehicle segmentation and position.Using local statistics to segment image which avoid vehicle edge loss caused by global threshold segmentation,then make use of projection method to locate vehicle.Besides,in this paper,it puts forward a location method for more than one cars in a lane,which lay the foundation for more vehicles recognition.(3) Vehicle feature extraction.According to requirements of the classification,this paper studies the vehicle’s shape feature which contain the length and height of vehicle,the apothem of car, the certain height in vehicle and moment characteristics of algebra which contain Hu invariant from edge image.(4) Vehicle recognition.In this paper,the vehicle models are divided cars,minibus, middle-size cars,trucks and large cars.Analysis the data of sample to select suitable feature to recognition.Using the MDC and KNN methods for vehicle classification,in addition,using MFC interface to display result of vehicle,find the reasons of error.