Research On 3D Measurment Of Large-scale Workpiece Based On Structured Light

With the rapid development of modern industrial manufacturing, a large number of parts are composed of the irregular surface; there is an urgent need for digital measurement of three-dimensional surface entities. Efficient and accurate 3D measurement technology during production process is required for modern manufacturing. As one of non-contact dimensional measurement methods, structured light 3D measurement technology has become a hot area of research. For the advantages of wide measurement range, high accuracy, easy extension, well adaptability and so on, it achieved many applications in cultural relic protection, industrial automation, reverse engineering and other fields.This paper revolves around the implementation of large-scale parts key dimension 3D measuring and its point cloud computing based on structured light system, and focuses reserch of vision system calibtation, stripe image process, encoding and decoding algorithm, multi-view point cloud process and registration, etc. The hardware and software of large-scale workpiece 3D measurement system have been built based on study of above.In this thesis, at first, the optical 3D measurement method is introduced briefly, and the classification and application of the method of structured light measurement is expounded. Based on study model of structured light, the calculation formula of measured point coordinate values is derived. Then research is mainly focused on camera and projector calibration technology, including analyzing the existing calibration algorithm and designing reasonable calibration scheme. For the camera calibration, plane-based calibration strategy is applied to determine the intrinsic and extrinsic parameters of the camera; the projector calibration algorithm based on local homography realized for high calibration precision. After analysis and comparison, forward and inverse gray code patterns is adopted as structured light coding strategy in this thesis. Because of the inter-reflection of object, the existied stripe boundary extraction algorithm have bad performance to get high precision decode result; then the invalid region extraction algorithm based on direct components and global components is realized. To solve the line equation of rotation axis, a novel method based on pose of planar calibration board is presented. By combining the decode information with system calibration parameters, 3D point cloud of measured object can be calculated. At last, a set of two-axis motion platform based on the movement control is designed to realize acquiring multi-view 3D point clouds automatically. The experiment results show that the design of the measuring system in this paper can realize the 3D measurement and reconstruction of the object applicably.