Research On 3D-Surface Non-Contacting Measurement And Registration Methodology

Laser scanning measurement which integrates electronics, computer vision technology, image processing technology etc. is a modern measuring technology. With it the three dimensional geometrical shape of the workiece can be measured without contacting the surface of workpiece, i.e. the numeric model of the workpiece is obtained quickly and accurately. The thesis aims to measure the three dimensional shape of the formed workpiece rapidly and calculate the shape error between the designed model and the workpiece. With the shape error feedback, the workpiece could be reformed to obtain high precision in Multi-Point forming process.Firstly the state of arts and developing trend of Laser scanning measurement technology are reviewed, and also the applications. The self developed Laser scanning measurement machine are introduced in detail, including the hardware and software, the working principles, the source of measuring error and the characteristic of data. As preprocessing measures, denoising and filtering methods are studied and realized with C++ language for the cloud data. And then the two step registration method combining ICP algorithm and Genetic Algorithm(GA) optimization is proposed. The main contents and results are as follows:1) To calculate the shape error between the objective model and formed workpiece, the shape is measured with the self developed non-contacting Laser scanning measurement machine. The inside and outside contour shape ofworkpiece can be measured without contacting the workpiece. The measurement is rapid, precision high and measured data uniform.2) The measured data is reconstructed with NURBS method. There will be large big error when the data is registred with objective model due to noises. Median filter is employed to improve the influence of noises. Then the surface denoised is output in IGES format to transfer data to other CAD software or MPF CAD software.3) To calculate the shape error of the workpiece in Multi-Point Forming, the two step registration method combining ICP algorithm and Genetic Algorithm(GA) optimization is proposed. At first step, the self adaptive GA is employed to aligning the data in different coordinate system, and then ICP algorithm is employed to register the data precisely by using the result of GA as the initial value. ICP algorithm, which can obtain registration result with high precision when the initial position of two models is appropriate. Although the result of the GA has error in a certain extent, it can assure two models that to be registered by ICP in an appropriate initial position.The paper bases on laser measurement technology, aims to analyze the forming error in MPF. A measure is proposed for error compensation in MPF by registration between two models. Experimentations results show that laser measuring technology and surface registration technology can be used in actual problems for 3D shape error analysis, closed loop compensation and reverse engineering in MPF.