The Study Of Vehicle Detection Based On HOG And SVM

Along with the rapidly developing of economics and the gradually increasing of living standard,the per capita retain number of automobile rises gradually.The traveling become more and more convenient and efficient.At the same time,every kind traffic problem is becoming more and more serious.For the effective traffic management,risking prevention,controlling the vehicle driving situation,to make the traffic unmanned intellectualized faster is around the corner.During this progress,the vehicle detection is step which is essential and necessary.Based on the study of some vehicle detection technologies,some algorithm in the field of machine learning and computer vision,giving the study on vehicle detection algorithm based on HOG and SVM.At first,do simple introduction about the background of the vehicle detection technology,given the purpose and meaning of the study of algorithm,telling the domestic and overseas study situation.After this,the related technology and algorithm,the used tools and framework,the open library during the process of the implements and study of this algorithm is also concluded.By telling the kind of vehicle detection technology,choose the video vehicle detection technology as the data source for the following algorithm study,and the meaning and advantages of this technology is also given.Then choose Histogram of Oriented Gradient extracting feature as the algorithm to extract the vehicle features by considering the applying situation and requests.Then give a simple introduction the basic theory of this extracting feature algorithm.Choose the Support Vector Machine as the classifier to detect and classify the related simple data,giving the basic theory of SVM.Telling how to use the Histogram of Oriented Gradient and Support Vector Machine together to implement the vehicle detection with the implementation plan.Aimed the problem that the Support Vector Machine may make itself overfitting during the training process,the improved plan and the detail steps were given.Consider the related theory and thinking about software engineering,Histogram of Oriented Gradient,Support Vector Machine and the improved plan,the program was implemented by using the Microsoft Company supplied MFC framework which is implemented by the C++ language and having the application program interface inside.The steps of the program during the implement process was wrote at great length,giving the testing and comparison aimed the implementation.After analyzed the detected result by several degree,the vehicle detection algorithm based on the HOG and SVM could able to giving the valid vehicle detected answer for related vehicle,which is helpful for the related problem and accurate.