Detection And Recognition Of The Front Vehicle Based On Monocular Video Flow

Intelligent vehicle in the social progress and the development of science and technology as a main topic,solved many difficult for the traditional vehicle security problem and innovation.Intelligent vehicle obstacle detection technology,therefore,more attention paid to the research,and reliably detect the front vehicle for intelligent driving has important research significance.This article is through the electric vehicle in front of the erection of the CCD camera to get the road information,including image processing,the lane line detection,interested in division,the HG-HV testing framework to identify the vehicles ahead.In front of the obstacle detection based on CCD camera can monitor electric car road environment,pose a potential safety hazard to electric detect obstacles.Paper main research content includes four aspects:1.The monocular video acquisition and image preprocessing.Introduced by the CCD camera in the video stream is through continuous static frame,and the need for image preprocessing before detection in front of the vehicle.Mainly includes the image pretreatment of image grayscale and binarization,image filtering,image edge detection,the selection based on Canny operator to detect image edges,experiments show that the second order differential operator in edge detection can effectively extract the image edge information.2.The extraction of interested area.The testing and identification of the vehicles ahead,you need to make sure the vehicles ahead of the candidate region,based on the improved Hough transform was carried out on the lane line identification and annotation,for region segmentation method with interested area,namely the interested region excluding to test the front of the vehicle,and the results verify that the system was divided by the lane line detection results.3.Vehicle detection identification.In the algorithm of front vehicle detection,this paper discusses the algorithm based on symmetry and the extraction of vehicle feature points,and focuses on the improvement of Harris corner detection in feature point extraction.It realizes breakthrough in real-time and innovates in removing boundary point detection,and as the core idea of the improved algorithm for the following symmetry and geometric characteristics of the vehicle phase to lay the foundation to ensure the accuracy of the vehicle test results.4.Based on the context of time and space of video vehicle tracking.After determining static frame of vehicle detection results,we need to track vehicle in video streaming,as essential to keep a safe distance part of the intelligent driving system.In this paper,in the context of time and space algorithm in vehicle tracking in video sequences,meeting over video sequence frame around the space context features.The front of the intelligent vehicle obstacle detection technology can effectively reduce the number of traffic accidents and to ensure the safety of vehicle driving,obstacle detection is not only applied to transportation,at the same time in industrial application,scientific exploration,disaster rescue,national defense military and other fields have a wide application prospect.