Research On Non-metric Camera Calibration And Photogrammetric Motion Measurement

Motion measurement based on non-metric cameras is to consecutively photograph the moving objects with cameras which are not professionally designed and manufactured for photogrammetry, and work out the important object motion parameters from images, such as position, attitude and velocity. Three difficulties lie in the process above: 1) the parameters of interior orientation model for the non-metric cameras are unstable, it brings a problem to interior orientation model calibrations; 2) the images of moving objects lack control information and that leads to difficult calibrations for exterior orientation model; 3) due to object’s small projection areas in images, it is a difficult work to estimate the motion parameters.Based on an overview of the photogrammetric methods for motion parameters at home and abroad, this dissertation studied the three difficult points above. The main works and important contributions of this dissertation are listed as follows:1. We concluded the basic flow of photogrammetric motion measurement based on non-metric cameras; then we specified the physical meanings of three independent rotation angles which are determined by the rotation matrix of a rigid body transform model; we built a general geometric imaging model for non-metric cameras which are suitable to both frame cameras and line scan cameras.2. For non-metric frame cameras and line scan cameras, one calibration method for interior orientation model was proposed respectively.(1) With a non-metric planar grid, a frame camera calibration method for interior orientation model was proposed to settle the problem of filling the camera’s large view range with a calibration pattern. This proposal skillfully makes use of the rotation matrix constraints and multi-view images of the same grid to work out the parameters of interior orientation model directly via a pair of degenerated projective equations. It is shown that the practical method can make use of the grid texture of man-made object to achieve accurate precision.(2) With the help of a linear stage, a new line scan camera calibration method for interior orientation model was proposed to address the problem that the information grabbed by line scan camera in static state is limited. In this method, three more extra motion parameters are introduced in the line scan camera model, and then many space points can be recognized in the scan image. Hence, primary orientation parameters are worked out from the dynamic line scan camera model. The experiments show that accurate calibration results can be achieved.3. For non-metric frame cameras and line scan cameras, one calibration method for exterior orientation model based on combined cameras was proposed respectively.(1) In order to cover moving objects in a large range, a photogrammetric camera system was designed by combining a measurement camera and a calibration camera together; and an automatic exterior orientation calibration method for the camera system was also proposed. The conjunct cameras are controlled by a synchronous controller, when the measurement camera swings to track a moving object; the calibration camera aims at the calibration scene and rotates along its optical axis. Therefore, the exterior orientation of measurement camera can be recovered indirectly from the images grabbed by the calibration camera. This system can be used to track and measure moving objects in large range with low cost.(2) A line scan camera calibration method was proposed by fixing to an auxiliary frame camera, and the calibration can be divided into two independent steps. Firstly, a 2D dynamic calibration scene is utilized to determine the elements of interior orientation of line scan camera and the relative position and orientation parameters of the two cameras; secondly, the exterior orientation model of the frame camera are firstly obtained via space resection, then the exterior orientation model are indirectly computed. The simulated and real data experiments show that the line scan camera calibration method based on combined cameras performs well.4. With object’s geometric information two solutions of motion parameters from images were proposed to address the problem that the object projection lacks information.(1) With several structured feature points, an estimation method of motion parameters was proposed to solve the projectile’s velocity, initial position and pose parameters from the stereo-scope line scan images. The proposal firstly builds an optimal mathematical model with imaging formulations; and then separates the pending parameters into linear ones and non-linear ones; secondly, Powell searching and linear least squares method are alternately used to solve the optimal matching model. It is validated by experimental results.(2) An automatic method for position and attitude measurement was proposed by tracking the planar region of a moving object with planar structures. In this method, the measurement results of the last frame are used to construct a planar template with object’s geometric and texture information; then the template is used to build and solve the optimal optimization model. Therefore, a recursive strategy for the image sequence measurement is implemented. This proposal promotes the intellectualized level of measurement.