Research On Videogrammetry For Large-Scale 3D Surface Based On Constraint Algorithm

The 3D measurement of large scale surfaces is a crucial technology for product quality control in heavy industry. Videogrammetry has the advantages of non-contact measurement and high-speed data acquisition, and it can be applied widely in large-scale 3D surface measurement. This paper aims to improve the videogrammetric accuracy for large-scale 3D measurement, focuses on large-scale 3D videogrammetry based on constraint algorithm.A new algorithm with the objective function, which incorporated the distance constraint to collinearity equations, is presented. This new objective function is adopted to reduce the effect of image distortion, and improves the accuracy of measurement. A test field, in which every marked point is measured by V-STARS (Simultanenous Triangulation and Resection System, the product of Geodetic System, Inc.), is built. The coordinates of the object points in the test field are computed by the bundle adjustment algorithm and the new algorithm respectively. The results show that the new algorithm is better than the bundle adjustment in precision.The theories deducing and implementation of multi-constraint algorithm is presented. For different scenes, the constraint condition varies correspondingly. Distance constraints algorithm is extended to collinear and co-plane constraints to improve the adaptability of videogrammetric system based on constraint algorithm. Kinds of constraints existed in scene can be introduced in the measurement system to improve accuracy.Camera calibration is the emphasis to videogrammetric system. Two calibration methods are studied: virtual grid calibration method and distance constraint calibration method. According to the property of the vanish point in affine geometry, the virtual grid calibration method is able to calibrate the principle point, distortion parameters and principle distance respectively. This method improves the calibration accuracy by decreasing the correlation of each parameter. Distance constraint calibration method calibrates the camera parameters through building the mathematic model between the coordinates of image points and the space distance. This method, locating the calibration constraint at the edge of the scene range, is more flexible than common used mehod.The matching between two images with near view points is completed by affine transformation. This image matching process is evaluated respectively by the indexes of wrong match rate, un-match rate and correct match rate. The results indicate that affine transformation is capable of the matching of images with near view points.Large-scale videogrammetric system based on the distance constraint is designed for the measurement of the surfaces of the gossamer inflatable antenna and deployable antenna. Largest marked residual and tightness are employed to assess the measurement accuracy of this videogrammetric system. The results of practical measurements show that this system is suitable for the 3D measurement of large-scale surface of the soft material. It is also a competitive choice for plenty of applications in measurement area.