Pavement Roughness Based On Binocular Vision Measurement System Research

The running speed of high-speed rail is high,and the unevenness of the roadbed has an important influence on its running stability and service life.The extraction of pavement roughness information can provide specific location and roughness information for pavement repair,which is of great significance for the later construction of high-speed rail.In this paper,a pavement roughness measurement system is designed and built,which realizes the measurement of pavement roughness.The experimental data show that the measurement accuracy of this system is high,which meets the needs of practical engineering.The specific work is as follows:(1)The principle of binocular vision ranging is studied,its accuracy and system structure are analyzed,the structural parameters of the binocular system are optimized,the experimental hardware is selected according to the actual working conditions,and a trolley for unevenness detection is built as the experimental platform.(2)Based on the principle of Zhang Zheng you’s calibration method,the C++language is used to call the OpenCV library programming to realize the automatic calibration of the binocular camera program to obtain the internal and external parameters and distortion coefficients of the camera.Through repeated experiments,the optimal parameters for the number of calibration images and the inclination angle of the calibration plate to improve the calibration accuracy were explored.Compared with the calibration results obtained by randomly collecting the images of the calibration plate,the calibration accuracy has been greatly improved.(3)The method of image matching and 3D ranging is studied,and the left and right camera images are matched and parallax calculated by semi-global block matching algorithm to realize 3D ranging,and the extraction of 3D information of road surface is completed.The C++ language is used to call the point cloud library programming,and after the three-dimensional point cloud of the road is preprocessed by noise reduction and sparseness,the random sampling consistency algorithm is used to fit the plane.The accuracy of distance measurement and flatness measurement of the system is verified by design experiments.The experimental data show that the system designed in this paper has high measurement accuracy and fitting plane accuracy,which can meet the actual engineering needs.(4)Aiming at the inconspicuous characteristics of road point clouds,a method of splicing road point clouds is proposed: place a certain number of marking discs in the public area of Singular value decomposition algorithm is used to solve the transformation matrix of front and rear road splicing,and the obtained transformation matrix is used as the initial value of NICP algorithm.Through continuous iteration and optimization,accurate matching is completed.The experimental data show that the measurement accuracy of unevenness after splicing meets the engineering practice.Finally,this paper uses Visual Studio 2019 as the platform,calls the OpenCV image library and point cloud library in C++,and implements various functional modules through programming,realizing the measurement of road roughness with binocular vision.At the same time,through the Microsoft basic class library programming,the completion Design of human-computer interface.