Three-dimensional Method For Measuring The Perpendicularity Of The Lower Wheelset Of Sparse Point Cloud

In this paper,based on the requirements of verticality measurement in the process of wheelset unloading,a wheelset unloading verticality measurement system is developed.The detection system uses linear laser displacement sensor,point laser displacement sensor,vision camera and other sensors,and uses laser tracker as auxiliary equipment to solve the problem of verticality measurement in the process of wheelset unloading.In this paper,surf algorithm is used to detect the image feature points,and find the feature relationship between the two images.The normal distribution is used to eliminate the unqualified corresponding points,and the improved longicorn whisker algorithm is used for image registration.The rotation and translation matrix between the left and right camera images is optimized,and the pixel coordinates of the image inside the registration image are calculated,Through the optimized rotation and translation matrix,the corresponding image pixel coordinates in the main image are obtained.This method improves the error of the conversion matrix in surf algorithm in the matching process.Finally,through the calibrated binocular camera internal and external parameters,the 3D point cloud reconstruction of the inner side of the wheel set is completed by using the epipolar method.In this paper,the improved bat algorithm is used to fit the 3D point cloud data collected by binocular camera.On the basis of bat algorithm,the adaptive threshold is added,that is,the algorithm of the same field is introduced.Firstly,the algorithm makes the initial calculation of the problem,uses the parameters calculated by the introduced algorithm as a reference,and changes it to the threshold of swarm intelligence algorithm(set precision / iteration times),which is called adaptive threshold here.This method has high precision and high efficiency,which avoids the bat algorithm falling into the local optimal problem to a certain extent,and changes the problem that it is not easy to set the iteration times and iteration accuracy in the previous intelligent algorithm optimization process,and finally completes the fitting of the inner side of the wheel.In this paper,a new cylinder fitting algorithm is proposed.Firstly,seven initial values of the cylinder are selected accurately,radius: r;Direction vector of cylinder axis: a,b,c;And the coordinates of any point that the axis passes through: x,y,z;By using plane fitting,circle fitting and three-dimensional data transformation from coordinate system to two-dimensional data,seven better initial values of the cylinder are found.Because these seven initial values are close to the seven real data of the cylinder,the combination of seven more accurate initial values and particle swarm optimization algorithm reduces a lot of iteration time in the subsequent iteration process,And accurately calculate the parameters of the cylinder,complete the optimization of cylinder fitting.That is to say,an initial value calculation method is introduced to replace it in the initialization process of swarm intelligence algorithm,so as to improve the iteration efficiency and avoid falling into local optimum.This paper analyzes the transformation relationship among binocular stereo vision coordinate system,point laser displacement sensor coordinate system and line laser displacement sensor coordinate system,and completes the transformation between each local coordinate system and global coordinate system through laser tracker: the transformation between line laser displacement sensor coordinate system and global coordinate system,the transformation between line laser displacement sensor coordinate system and global coordinate system The conversion between the coordinate system of the point laser displacement sensor and the coordinate system of the binocular camera,and the conversion between the coordinate system of the binocular camera and the global coordinate system can convert all the measured data into the same coordinate system,which is convenient for calculation.