| Railway track is the guarantee of high-speed operation of the train.In the course of long-term operation,the geometrical deformation of track may occur,and the geometrical detection and maintenance of the railway track must be strengthened.In this thesis,through the close-range photogramme try technique for the analysis of long distance orbit close-range image,a set of orbit image adjustment algorithm is explored and exterior orientation element and encryption point coordinates of orbit image is accurately calculated,which provides data for geometry detection of railway track.Studying on the coefficient matrix of the orbital close-range image,it is found that the coefficient matrix of the error equation and the law equation is sparse,and the distribution is regular.According to the unique data structure characteristics of orbit image,using multi-wave front method based on UMFPACK solver for large sparse matrix storage,operation efficiency is improved and memory is less occupied.The problem of large sparse matrix data storage is overcome,and it is suitable for the storage and calculation of large sparse data in long distance orbit close range image.In order to avoid the fluctuation caused by single track adjustment in the process of long distance orbit image adjustment,this thesisr adopts the method of double-track adjustment and two-track combined adjustment,forming a multi-baseline strip network,and analyzes the problem of camera distortion in the adjustment process,exploring correct distortion model.Then through the image relative orientation,model connection and absolute orientation,it provides initial value for block bundle adjustment.Due to the large number of images and high similarity of adjacent image grey level and the less quantity of control points,it will be a case of low accuracy when we stay away from the control points in the adjustment process.We propose a block based sparse matrix compression(SMC)technique combined with the pretreatment conjugate gradient method,making sparse matrix easy to be updated and improving the convergence rate of adjustment.This thesis introduces ill-posedness problem in the adjustment and the corresponding processing method of long distance subsection adjustment to ensure the stability achievement of adjustment.A set of adjustment algorithm is designed for the long distance track close-range image data.Taking close-range image of Ziyang railway section as experimental data,through the multi wave front data structure storage,image cross joint adjustment and model connection,an initial data is brought into the conjugate gradient bundle adjustment method.Exterior orientation elements of orbital image is calculated correctly and then it provides data for calculating geometric parameter of railway track. |
