Research On Vehicle Type Recognition Method Of Road Vehicles

With the improvement of living standards, the number of private cars increasesevery year. Manual management is difficult to meet traffic intelligence managementrequirements, including traffic detection, management fees, vehicle identification, etc.In recent years, due to the rapid development of computer vision and patternrecognition technology, vehicle classification technology based on the video imagehas been widely concerned in traffic control. This paper researched car modelsrecognition in depth, the process of vehicle recognition has three steps: movingvehicles detecting, feature extraction and vehicle recognition.This paper described commonly method of image segmentation and proposed anadaptive threshold method based on histogram information. Typical simple movingtarget detection method was described, using a modified Gaussian mixture model todetect the moving vehicles. This paper removed the shadow information according tothe gray image characteristics of the vehicle, then got a complete vehicle targetbinding region notation.For the vehicle models feature extraction and recognition, this paper introducedcommonly methods firstly. Through processing of vehicle target detection area, thispaper proposed a area projection for statistical traffic, the method can accurately countthe number of vehicles; and calculate average size of the detection area of the vehicle,use the K- means clustering algorithm for model classification.In this paper, MATLAB7.1 programming software was used to develop thevehicle recognition system, and the experimental data and results showed that theproposed traffic statistics and identification method has a strong feasibility andpracticality.