Research On Key Techniques Of Large Scale Dynamic Photogrammetry

With the development of scientific and industrial techniques,the demand for large and ultra-large equipment is growing.In order to guarantee the quality of these equipment,it is of great significance to study and develop measuring techniques for large scale and dynamic applications.For its advantages in noncontact,high accuracy and multi point measurement,photogrammetry has become an important approach for measuring geometric quantities.However,in large scale dynamic applications,new problems and difficulties arise,which reduce the accuracy,increase complexity of measurement operation and data analysis and thus block the engineering applications of dynamic photogrammetry in large scale volume.It is of great urgency to study new theories and methods of photogrammetry that aim the problems and difficulties in large scale dynamic measurement applications.In order to solve the problems and overcome the difficulties of large scale dynamic photogrammetry,the model and calibration method of camera distortion,the calibration and evaluation method of large scale dynamic systems in-situ and the analyzing method of deformation data are studied.Additionally,a large amount of experiments and engineering applications are conducted to validate the research works which include:(1)In order to model the distortion mechanism of a principle-distance-invariable photogrammetric camera in large scale scenes,the depth-dependent radial and decentering distortion model and computation method are studied by establishing the relationship between the focusing and defocusing distortion parameters and meanwhile introducing the magnification-dependent distortion model.The calibrating method of distortion parameters based on coplanar line patterns are proposed.Additionally,the design of calibration object,measuring and adjusting of camera attitude,image processing and block adjustment model of distortion parameters based on linear patterns are studied,by which high accurate,efficient and automated camera calibration is achieved.Experiment results show that,variation of object depth brings non-negligible changes in distortion parameters for a principle-distance-invariable camera,that the effect of object depth on distortion is reduced significantly by the proposed distortion model,and that in a volume of 7.5m(length)×3.5m(height)×2.5m(depth)the large scale spatial measurement precision is improved by 40%compared with the constant parameter distortion model.(2)In order to reduce the complexity of calibration operation and improve measurement precision of large scale dynamic photogrammetric systems,a simultaneous all parameter calibrating and orientating method of multi camera systems based on a moving scale bar is proposed.The model of relative exterior orientation,the estimation of relative exterior orientation parameters and the model and block computation of self-calibrating bundle adjustment that is constrained by spatial point-to-point distances are studied and elaborated.The internal precision based on error propagation and the external accuracy based on reconstruction errors of spatial lengths are proposed to assess calibration performance.The method is validated by a large amount of simulations and real data experiments.The method achieves an extended relative precision of 1/18000(k=3)in large measurement volume,and improves spatial measurement precision evidently compared with the traditional 3D point array self-calibration method.(3)To measure 3D deformation of a moving object,a deformation analyzing method is propose for a category of objects that have an explicit mathematical description.The deformation of moving objects is modeled by residual errors of rigid body transformation and the mathematical function involving deformation,the measurement coordinates of object points and transformation parameters is established analytically.Additionally,the numerical computing method of optimal transformation parameters that minimize the sum of squares of deformation is proposed.The method realizes the best fit of measurement points to the model of the measured object and helps to obtain the deformation of the points away from the designed model.The model and calculation method are applied in structural measurement and regulation of large parabolic trough solar concentrators and achieve high accurate information of positional and angular errors.Compared with traditional static photogrammetric method,the proposed method significantly improves regulation efficiency and the geometric accuracy of the regulated structures.Additionally,the modal and method are applied in dynamic measurement of a 2.4m diameter parabolic antenna deformation.It is exhibited that high accurate deformation of the antenna surface under different elevation angles and revolution-per-minutes are achieved without construction of a reference system.