Development Of Mapper Of The Traffic Accident Scene Based On Binocular Stereo Omni-directional Vision Sensors

Nowadays, the manner of manual operation such as judgment by human eyes,tape measurement and manual drawing is still the main method to investigate theaccident scene in home and abroad. But this approach has some shortcomings inmeasurement system, such as low efficiency, low accuracy and high rate of omission.After the scene is damaged, it also cannot be surveyed again in the manner. Digitalcameras are used nowadays in the investigation of accident scenes, which are alsolimited by the image recording of the scene of the accident. It can’t get exact distanceand azimuth information of measurement points, so it never achieves the goal ofaccurate mapping.Binocular stereo vision is a computer vision technology to obtain depthinformation according to the parallax principle. Binocular stereo vision has obviousadvantages in the measurement, such as rapidity and reproducibility, so it is apromising application technology for surveying the accident scene.The main work of this paper has the following aspects:(1) On the basis of analyzing the characteristics of the accident scene mapping,this paper designed a Binocular Omni-Directional Vision Sensors(BODVS) that hasmany advantages, such as wide range of measurement, high accuracy and positioningaccuracy. BODVS is combined with GPS positioning, GIS map search and mappingsystems to form a new accident scene mapping system based on binocular stereopanoramic vision.(2) There are many problems in the measurement of binocular stereo vision anddrawing the traffic accident scene diagram, such as technology of the accuratecalibration and positioning graphics symbols. This paper presents an improvedcalibration method based on polynomially simplifying the calibration model for the catadioptric panoramic camera after researching existing calibration technologies,which can correct the calibration error caused by the non-aligned of mirror-camerasystem and camera lens distortion. The graphical symbol positioning technologycombines with measurement and drawing organically, so we can draw the accidentscene diagram automatically based on the measurement results.(3) This paper designs hundreds of the graphical symbols of accident sceneelements based on traffic accident scene drawing standards and the characteristics ofaccident scene measurement by BODVS. The graphical symbols can not only bedrawn manually by adding, modifying or deleting, but also be drawn automaticallybased on BODVS measurement results.(4) This paper utilizes epipolar lines constraints and the laser radial strongconstraints to solve mismatching problem and the problem of low light in measuringat night by binocular stereo vision.