Research On Cross-Ratio Based Camera Calibration And Vibration Correction In Digital Streak Photogrammetric Measurement

Moving parameter measurement is an important problem in close-range photogrammetry and computer vision. Optical image based moving parameter measurement technique has theoretical support from classic photogrammetry and computer vision. Compared with traditional radar, laser and GPS based measurement techniques, this technique has big advantages due to the characteristics of uncontact measurement and high accuracy.Line scan cameras are widely used in space-borne remote sensing areas as a new sensor. Line scan cameras also get big improvements in close-range photogrammetry since recent years. Compared with frame cameras, line scan cameras have higher frame rate and space resolution and thus are replacing frame cameras in many moving parameter measurement applications. Digital streak photogrammetric measurement system is a shooting range system which adopts line scan cameras as imaging sensors. This system can obtain the stereo scan images of a pill in two angles of view with a mirror and then measures the speed and attack angle of this pill. The frame rate of line scan cameras in this system can reach tens of thousands of frames per second and thus has a special advantage in moving parameter measurement of high-speed pills. On the other hand, compared with two dimensional pictures, linear pictures contain so little geometric information that bring two difficulties to digital streak photogrammetric measurement system: first, calibration of intrinsic parameters of line scan cameras with linear pictures; second, calibration of relative translation and rotation parameters between the line scan camera and the mirror; third, vibration correction with linear pictures.In this dissertation, space point coordinates are calculated from linear pictures using the invariant of cross-ratio based on the geometric imaging model of line scan cameras. A research of the above three difficulties is proposed. The main works and important contributions of this dissertation are listed as follows:1. A cross-ratio invariant based line scan camera calibration method is proposed for the geometrical line scan camera imaging model. Compared with traditional calibration methods, this method does not require translation of the three dimensional calibration pattern along a certain direction. In this method, special linear features are drawn on the calibration pattern. Then, the space intersection points of the view plane of the line scan camera with linear features are calculated based on the invariant of cross-ratio. With a traditional two-step calibration method, the intrinsic parameters of the line scan camera are solved from the image points and their corresponding space intersection points. The experiments show that stable and accurate calibration results can be achieved.2. A calibration method is proposed for the relative translation and rotation parameters between the line scan camera and the mirror. This method includes three steps:(1)First, the line scan camera requiring calibration is coupled with a calibrated frame camera rigidly. Then, several views of the planar calibration pattern under different orientations are captured by the line scan camera and frame camera. The rigid transform parameters between the two cameras are solved with the rotation matrix constraints.(2)Using the markers on the mirror, relative translation and rotation parameters between the frame camera and mirror can be solved directly(3)The relative translation and rotation parameters between the line scan camera and the mirror can be solved indirectly.3. A vibration correction method is proposed for the mirror. In this method, linear features are drawn on the upper and lower border of the mirror. The position and orientation of the mirror are solved from each frame based on the invariant of the cross-ratio. Then the linear scan image is rectified to minish the moving parameter measurement error due to the vibration of the mirror.