Research On Forward Vehicle Speed Identification Based On Image Perception Technology

With the rapid development of the national economy,the number of motor vehicles in China continues to increase,coupled with imperfect road infrastructure,the quality of drivers is uneven,the road safety situation is becoming more and more serious,traffic safety accidents are frequent,and in the process of traffic accidents,The speed of the vehicle before the accident is of great significance.In recent years,the popular installation of on-board driving video recorders and driving video have become important image information in traffic accidents.Therefore,how to use the driving video to obtain the running speed of the preceding vehicles has urgent social demand and broad application prospects.The traditional traffic accident video speed calculation mainly includes the methods based on vehicle reference object and based on ground reference object.This thesis proposes a forward vehicle video speed measurement method,which can be used as a useful supplement to the existing method.Using the driving video recorder to obtain the motion state of the vehicle in front,the computer vision technology,camera calibration technology and image information sensing technology are used to realize the identification of the running speed of the vehicle ahead.The main research contents are as follows:Aiming at the accurate perception problem of the key information of the vehicle's operating environment,the Canny edge detection of adaptive threshold is realized by using the image gradient distribution information.The road vanishing point is introduced to establish the space constraint of the edge line of the lane line in the Hough accumulating domain,and then the lane line parameters are obtained.A hypothetical and verified vehicle positioning method is proposed.The vehicle presence hypothesis area is generated by segmenting the vehicle bottom shadow,and the Haar-like feature training vehicle cascade classifier performs hypothetical area verification,which can effectively improve the positioning accuracy of the vehicle position.Aiming at the calibration problem outside the camera,based on the coordinate transformation and imaging principle,an external parameter self-calibration method is proposed.This method can be completed while the vehicle is traveling without prior calibration.Compared with the traditional scheme,considering the camera direction angle factor,it is more suitable for the situation where the camera installation position is uncontrollable in traffic accidents.The geometric model is used to measure the distance of the vehicle ahead,and the relative speed between the front vehicle and the vehicle is further obtained.Accurate identification of the running speed information of the vehicle ahead.Based on the above research,the front vehicle video speed measurement system was developed on the VS2010+OpenCV2.4.3 platform,and the running speed of the vehicle in front was calculated.The corresponding vehicle speed measurement experiment was designed to verify the accuracy of the algorithm.In addition,for the existing traffic accident video speed calculation scheme,the vehicle speed measurement experiment based on the vehicle body reference object and the ground reference object was carried out respectively.Finally,the error analysis of the experimental results was carried out.