In recent years, along with the increasing of city vehicles, in order to reduce thepressure on ground transportation, a large number of viaducts and culverts are built. Onone hand, many trucks are intentionally overloaded and over-height for reducing transportcost, in addition, due to environmental and other factors, the high limit signs becomeblurred and the height of the bridges and culverts in some city regions is lower than thereal standards. On the other hand, some vehicles should be over-height sometimes due tothe special transport demand When the drivers drive these over-height vehicles into thecity, they cannot predict the real height restriction of these bridges and have flukypsychology, which will result in frequent crash accidents.This paper studied the traffic capacity when the over-height vehicles went throughseveral typical restricted conditions based on on-board binocular stereo vision system.Firstly, this paper summarized the on-board binocular stereo vision system, completed thestereo camera calibration and got the internal and external parameters of stereo camera,then verified the correctness of these parameters and erected a stereo camera model.Secondly, this paper focused on the geometric information measurement method of threetypical kinds of bridges and tunnels. The upper part of images was set to interested area,and tunnels’ lower edge, the targets which were divided into linear, small arc type andsemi-circular according to the characteristics of the bridges, were recognized thoughHough transform and the method of curve fitting, then detected the Harris corners andmatched the feature point in the left and right images, got the three-dimensionalinformation of the matching points. Finally, this paper proposed risk estimation and alarmdecision making based on the early warning indicators and warning limit standard. Afterexperimental verification, the algorithm in this paper had good rationality and feasibility,can effectively reduce accidents of over-height vehicles’ crashing on bridges and providetheory guidance and technical support for intelligent vehicle and transportation security. |