Research On Calibration Problem Of Urban Roads And Expressways

Traffic monitoring video can obtain valuable traffic parameters,such as vehicle speed and vehicle three-dimensional information.After analysis,processing and effective use,it will have an important impact on areas such as smart transportation and autonomous driving.The premise is that the surveillance camera needs to be calibrated by the camera,and the internal and external parameters of the camera are used to establish the mapping relationship between image information and spatial information to obtain traffic parameters.The current camera self-calibration algorithm based on the vanishing point in traffic scenes has certain limitations:for example,there must be specific environments such as zebra crossings,lane lines,and buildings in the scene to obtain a straight line to obtain the vanishing point;the detection of the calibration object is affected by weather factors As a result,the detection accuracy is not high;the detection is difficult when the vanishing point tends to infinity.Based on the above problems,this paper uses the moving targets in the traffic scene to extract the vanishing points,and uses the camera self-calibration model based on three vanishing points to solve the internal and external parameters of the camera.The research content is as follows:1.The method of combining Gaussian mixture background modeling and homomorphic filtering algorithm is used for moving target detection.Gaussian modeling is performed on the pixel reflectivity value and the difference is obtained to obtain the foreground of the moving target.This method can effectively remove the light and shadow in the scene and improve the detection rate of the moving target.And through the moving target classification method proposed in this paper,the required moving targets,namely vehicles and pedestrians,are obtained.2.Extraction of line equations in urban road and highway scenes.Because it is difficult to obtain the vanishing point in the traffic scene,this article uses the moving target in the scene to obtain the vanishing point.At the same time,considering the characteristics of different scenes,combining the extracted moving targets,different algorithms are used to accurately obtain the lines in the two scenes.equation.3.Use the cascaded Hough transform algorithm to estimate the vanishing point.Drawing lessons from the traditional Hough transform idea,a diamond-shaped space-based cascaded Hough transform is established to extract the vanishing points in the scene.This algorithm effectively solves the problem of difficult detection when the vanishing point is infinite,and detects the position of the vanishing point more stably and accurately.The algorithm in this paper has undergone a large number of experimental tests in the two given traffic scenarios.And the experimental results show that: in the scenes of urban roads and expressways,the average errors between the calibrated line segments length and the known line segments length are 5.54% and 5.38%,this paper also analyzes the reasons for the errors,it can be seen that the algorithm has high accuracy and wide applicability.