Research On Key Technology Based On Line Structured Light Vision Measurement

Line structured light 3D vision measurement is a three-dimensional measurement technique based on laser triangulation, which is one of the main methods to obtain the 3D data of the measured object. The technique is in fact an improvement and extension of the traditional stereo vision system, which can realize the accurate and efficient acquisition of 3D information on the surface of the measured model. The experimental platform is designed for line structure light 3D measurement in this paper, and the camera calibration of the line structure light 3D scanning system and light stripe center extraction are studied, finally, the data reconstruction technology based on 3D point cloud registration is researched.(1) The test platform for line structured light 3D scanning system is built. The test bench has the advantages of simple structure, low cost, easy to operate, easy to fitting and dismantiling. Through the rotation and movement of the experimental platform we can complete more complex object or a larger volume of the object surface 3D vision measurement.(2)Three different camera calibration methods based on two dimensional plane target are studied. Through the different calibration methods using the two dimensional target, a higher accuracy of the camera calibration is achieved, and the merits of the three calibration methods are evaluated. The camera calibration method integrated R2013b Matlab and OpenCV software is proposed, the camera calibration, the calibration process is simple and easy to operate, the calibration accuracy is higher than the single calibration method.(3) The methods for extracting the center of light stripe are discussed in detail, and the center of light stripe through the self-made sphere model are extracted,then single field of light stripe of 3D point cloud information are achieved in the camera coordinate system. The typical 3D point cloud registration method is discussed and the rotating platform rotating different fields of 3D point cloud data, based on arbitrary rotation angle of the rotating platform, different fields of point cloud are stitched based on marked point stitching technique using the Least-Square method.(4) Self-made markers are pasted on the free surface model and rat type accumulator. From different perspectives of the model, combined with the Least-square method, different perspectives of the surface point cloud registration are verified. Then point cloud data are fitted and reconstructed and eviation comparison, finally, the error of the point cloud registration process is analyzed.