The Optimal Method Research For The Position Measurement Of Optical Fibers

With the continuous development of the multi-object fiber spectroscopic telescope, the integrated number of optical fibers on the focal plane is increasing, from400to4000, and even5000in the future. So far, the largest multi-object fiber spectroscopic telescope is the large area multi-object fiber spectroscopy telescope, namely LAMOST. There are4000optical fibers on the focal plane of LAMOST, which are employed to track the celestial targets accurately. With the increase of the fiber number, the fiber positioning can’t be accomplished artificially anymore. Aiming to control the fiber tips on the focal surface of the telescope, the movement trajectories of the fiber tips must be measured in advance, so the crucial problem is the fiber position measurement. Due to the special construction of the large-area focal plane and the small diameter of the fiber, only the vision measurement method featured with high precision and high efficiency meets the requirement.This paper has analyzed the geometric model and the calibration method of the camera. Based on the principle of the visual measurement, the key factors has been researched and improved. The bundle adjustment and the improved method have been studied in details. Then the simulations and real experiments have been carried out for verification.The camera is applied as measuring tool to analyze the linear and non-linear model of the image by ping hole. In view of the aberration model and the calibration method of the camera, an optimal method for extracting the elliptical feature point has been developed. The camera calibration experiment based on the circular calibration board has been executed to gain the calibration results and reprojection errors. The experiment results prove that the fine extract method with elliptical feature point is of comparable higher precision.Based on the comparison of various algorithms for vision measurement, this paper presents a new measurement method featured with high precision and high efficiency for LAMOST. A polynomial fitting method suitable for plane measurement based on the geometric model of camera has been researched. The mapping relationship between the coordinate of the image points and the XY coordinate of the3D points. The experiment result indicates that the binocular vision and the polynomial fitting method is inadaptable for the large scale measurement. There are many key factors affecting the measure accuracy, such as the optical fiber source, the light intensity, and the emergence angle. The optimization of these factors has been researched in details. The bundle adjustment is employed to optimize the parameters of the vision measurement. The bundle adjustment and the LM non-linear iterative method are brought for measurement. The normal equations are applied to sparse bundle adjustment. This method can save more calculation time and memory space. Since the normal bundle adjustment can’t reach the accuracy requirement, an improved bundle adjustment method with only one high-precision control point has been developed. The results of both the simulation and the real experiment indicate the new algorithm is feasible and it’s maximum relative accuracy is as high as1/40000.