Research On Vehicle Type Recognition Technology In Intelligent Traffic Surveillance Video

In recent years,Intelligent Transportation System(ITS)is the development direction of future traffic.Vehicle type recognition plays a more important role in both traffic planning and road monitoring with the increasing number of vehicles,with great research and application value.There are few differences in the characteristics of vehicles and many interference factors in vehicle identification,especially in complex backgrounds.In order to improve the accuracy of vehicle detection and recognition in traffic scenes,a vehicle-types recognition technology based on computer vision is proposed in this paper.In this paper,traffic road video image is taken as the research object.In order to eliminate the interference and noise of complex external environment,a preprocessing scheme including image denoising and image enhancement is adopted.Then in the vehicle detection part of the video sequence,this paper analyzes three algorithms commonly used moving target detection in detail.According to the difference of the moving speed of the vehicle target and the interference of the light intensity change,the background-difference method based on the mixed Gaussian model is proposed to identify the moving vehicle,eliminating the phenomenon of “shadow” and “cavity” and avoiding the influence of illumination factors on the detection results;According to the obtained moving target detection result,since only the binarized vehicle contour is included and there is not enough texture,edge and other local information,the above-mentioned binarization result is sequentially subjected to morphological operations,connected-domain analysis,the Graphcut method,and the adaptive bounding box completely separate the foreground objects containing the vehicle information from the original image.The LBP and HOG feature is then introduced to obtain rich features of the image,and the SVM classifier in machine learning is trained at the output layer by multitasking learning of a large amount of tagged data.Different from the traditional method,the PCA dimension reduction process is used to speed up the improved HOG feature,the least squares method is used to avoid the classifier falling into the local minimum value,and the least square method is used to integrate the objective function solving process of support vector machine,and the optimal recognition effect is achieved by designing the parallel classifier structure.In this paper,the public vehicle data set is used as the classifier training set,including 9850 high-resolution vehicle front view images.The experimental results on the test data set and the common data set verify the proposed method and can better perform video images.Vehicle detection and vehicle identification.