Research On Vehicle Image Recognition Method Based On KPCA And Gaussian Process In Complex Environment

With the development of society and economy and the improvement of people's living standard,the number of vehicles is increasing and the penetration rate is getting higher and higher.The backward artificial transportation management can not meet the demand of modern traffic management.In recent years,the emergence of intelligent transportation system has made great contribution to traffic management.The goal of intelligent transportation system is to improve traffic safety and reduce the incidence of traffic accidents by detecting the surrounding environment and the basic and important research direction in intelligent traffic management is vehicle target recognition.Domestic and foreign scholars have successfully developed some methods of vehicle target recognition.However,there are still many shortcomings.For example,the accuracy of vehicle target recognition is not satisfactory in bad weather and other complex environments.It has become a hot research at home and aboard for the past few years.In this paper,the above problems were studied,a vehicle identification model based on KPCA complex environment is established.The model includes feature extraction and classifier construction.The feature extraction stage mainly uses KPCA to extract the feature descriptors of vehicle images.The classifier construction stage mainly includes the construction of support vector machine classifier and GPC classifier.And the algorithm advantage is verified through the comparison of experiments.Three problems are studied in this paper: firstly,in the stage of image preprocessing,an image adaptive enhancement algorithm is proposed to improve the image quality of vehicles in the environment mentioned above,aiming at the low visibility scene such as fog weather and night time.In addition,the gradient histogram feature extraction of vehicle image is realized.Secondly,aiming at the problem of high feature dimension of gradient histogram of vehicle image,which affects the efficiency of subsequent algorithm,this paper discusses the method of feature reduction based on principal component analysis(PCA)and kernel principal component analysis(KPCA).Thirdly,the vehicle classifier based on Gaussian random process is established.The above model is used in vehicle image recognition,which improves vehicle target recognition rate in complex environment.The experimental results show that the algorithm is greatly practical.The recognition rate of the vehicle targets is 98.7% in the complex environment.The method of vehicle target recognition proposed in this paper has achieved satisfactory results.It has great research significance and use value.