Research Of Preceding Moving Vehicles Detection And Distance Measurement For Intelligent Vehicle Based On Vision

The issue from security technology project of Shanxi Administration ofWork Safety Research and Design of Intelligent Monitoring System of FatigueDriving, the previous seniors have researched the Monitoring System ofFatigue Driving from drivers point of view, and the fatigue status of drivers isbeing constantly monitored and warned to ensure driving safety. However, theaccident is still high due to the uncertainty of human factors. And the accidentscan be effectively reduced by intelligent vehicle technology, therefore thistechnology attracts more and more attention. Based on the above reasons, thedetection technologies and distance measurement methods of the movingvehicle in front of the intelligent vehicle are studied in this paper. The detectionand distance measurement for the moving vehicle are realized throughMATLAB programs and experiments. For the above studies, the main contentsof this thesis are as follows:(1)This paper designs an effective vehicle detection scheme to study thedetection technologies of the moving vehicle in front of the intelligent vehicle.Firstly, the project analyzes preprocessing techniques of the road imagecomprehensively and selects the proper preprocessing algorithm by comparing the experiment results. And it realizes the lane detection with improved Houghtransform algorithm under the assumption of linear roads model and delineatesthe effective road surface region which narrows the detection area. Finally, itachieves a preliminary detection for the vehicle according to the shadows underthe forward vehicle and establishes an interest region, verifying it by utilizingthe texture and gray symmetry characteristics of the vehicle to eliminate theinterference of non-vehicle factors completely and detect the moving vehicle infront of the intelligent vehicle accurately. The effectiveness of the vehicledetection scheme is verified by comparing with large numbers of experiments.(2)This paper studies the distance measurement methods between theintelligent vehicle and the forward vehicle. The article orientates the planar pointin space through geometrical deduction based on the principle of visualprojection model and establishes the transformational relation between the roadplane coordinate system (the world coordinate system) and the pixel coordinatesystem. Then it combined with the calibration results of intrinsic parameters ofcamera. Finally, the distance between two vehicles is calculated by the distancecalculation formula. By comparing the experimental measurement, the accuracyof location method is proved based on monocular vision.