Research On Key Technologies Of 3D Measurement System Based On Laser Triangulation

The laser triangulation is a principle simply,measurement accuracy highly and real-time strongly non-contact 3D measurement technology,however,compared with foreign countries,the domestic laser triangulation technology has a certain gap in accuracy,real-time and flexibility.For the above questions,this paper based on the analysis of how to design and construct the laser triangle system,the key technologies of high precision laser triangulation system,such as efficient calibration,light strip center feature extraction and point cloud registration,are researched and improved,and the function of these technologies and the measurement accuracy of the system applied to actual workpiece measurement are verified by relevant experiments.In this paper,the theory of laser trigonometry is first analyzed theoretically,and the system structure design and component choice are completed on the premise of definite design parameters and indexes.Next,the relevant principles and methods of system calibration are investigated.So as to ensure the light plane pose precision,an optimized light plane calibration way is proposed,which is mainly in view of adaptive threshold value.System calibration experiments show that the overall average error is 0.0685352 pixels of the camera,and the residual value of optical plane fitting is 4.85575e^-5,the accuracy is high,and the optical plane can be perfectly fitted.In order to improve the algorithm accuracy of the light strip center feature extraction,an improved method combining Zhang-Suen thinning algorithm and Hessian matrix is proposed in combination with the actual characteristics of the system strip image.The experiments indicate that the optimized algorithm is thinner and more accurate on the laser lines extracted than the gray center of method and Steger method while reducing the amount of computation.Then the laser triangulation system is used to measure the height of workpieces.The experimental datas show that the absolute measurement error of the two workpieces is less than 0.1mm and the relative error is less than 2%.The reconstruction measurement of the workpiece can be realized while ensuring the accuracy.Aiming at the limitations of single view 3D measurement,this paper studies the point cloud registration algorithm in point cloud mosaic technology under multi view measurement ased based on feature matching.Because the problems of high false matching rate and long matching time when the traditional feature matching algorithm changes the image rotation scale,an improved feature matching algorithm based on local hold matching is proposed.The experimental results show that the matching accuracy of the algorithm can reach more than95%,the removal rate of false matching is high,and the average matching time is shortened by 0.83s,and the registration process of point cloud can be effectively realized,which provides precision support for the subsequent point cloud stitching technology in the case of multiple perspectives.