| With the development of science and technology,the pose measurement technology of space target plays an important role in the fields of aviation,aerospace,industry and so on.The pose measurement technology based on optical image is a research hotspot in the fields of close range photogrammetry,computer vision and remote sensing,which has the advantages of non-contact,simple equipment and high measurement accuracy.In recent years,with the development of linear CCD,COMS sensor technology and the in-depth study of linear optical image,the advantages of linear optical image sensor in space target pose measurement are shown.Compared with other non-contact measurement methods such as laser and radar,it has an irreplaceable role.Aiming at the contradiction between the accuracy and speed of traditional pose measurement technology based on area array optical image,this paper presents a high precision pose estimation method for spatial objects based on one-dimensional image combination,which is intended to complete the high-precision pose measurement task of attitude spatial floating moving target.According to the characteristics of linear array signal,an improved method of high-precision image point centroid positioning is proposed.Based on the imaging mechanism of linear array camera,a new calibration method of linear array camera is studied.Based on the measurement principle of multiple linear array camera,the mathematical model of multi linear array camera pose measurement system is established,The improved algorithm of pose estimation for space target based on multi linear array combined measurement is studied,and a high-precision and fast pose calculation model for cooperative target is built.The method proposed in this paper can verify the motion parameters of the target with INS.Therefore,it will have a wide application prospect to realize the motion detection of space moving point target by multi linear array CCD and measure the pose of the moving target with high preci sion.The main research work of this paper include the following parts:(1)A pose measurement system composed of three line array CCD is designed and developed.The factors that affect the change of the common field of view of the three line array camera are studied,and the effective working field of view space of the camera is analyzed.Based on the original linear array signal,according to the distribution of noise,an improved wavelet threshold denoising method is studied.The results show that the algorithm can suppress all kinds of noise.In order to improve the image point positioning ability of linear CCD,aiming at the shortcomings of traditional one-dimensional signal image point positioning method,an interpolation barycenter weighted centroid positioning method with adaptive threshold is proposed.The research shows that the centroid positioning error of this algorithm is within 0.029 pixels.Compared with other subdivision methods,this algorithm has higher positioning accuracy and better anti noise performance,which can meet the task requirements of high-precision system.(2)Aiming at the distortion problem of linear camera equipped with cylindrical lens in pose measurement,a modified mathematical model of linear camera distortion based on constant cross ratio is established,and an improved two-step camera calibration method based on linear CCD is proposed.This method does not need to calibrate the reference object with high precision,but only uses the constraint relationship between the targets to establish the distortion model of linear array camera based on the invariable characteristic of cross ratio.Firstly,the linear solution of the camera parameters of the established linear array imaging model is carried out.Then the distortion coefficient is calculated.Finally,the nonlinear optimization of the image parameters is conducted out by the iterative method to complete the calibration of the linear array camera.The simulation results show that compared with the traditional DLT method,the algorithm in this paper has better anti noise performance and higher accuracy.The results show that the main point error has a greater impact on the pose measurement accuracy than the equivalent focal length.Finally,the re-projection standard deviation of this algorithm is 0.031 pixel.In the practical verification,firstly,aiming at the distortion model and the calibration accuracy of the single camera,the re-projection positioning error is better than 0.025 pixel,In the calibration scheme,the inherent pose relationship between linear cameras is introduced as the constraint condition of calibration error compensation,which is added to the collinear error equation of linear camera calibration.After compensation,the position error is within 0.746 mm and the angle error is within 0.247 °,compared with the traditional linear camera calibration method,this method has higher calibration accuracy and stronger stability.(3)The traditional orthogonal iterative algorithm is based on monocular visio n measurement,and it is only applied to the area array camera.In order to solve the pose estimation problem of multi-linear array cameras,this paper proposes an high precision solution algorithm based on the pose estimation of multi linear array CCD.Firstly,a unified representation of multi-camera data is established,which regards all camera observation data as data under the generalized single camera.Then,a new error evaluation function by introducing image residual as the weight judgment criterion is built.Finally,the improved weighted orthogonal iterative algorithm is applied to solve the pose parameters.The proposed algorithm effectively overcomes the influence of errors of outliers on method accuracy in the traditional orthogonal iterative algorithm.The simulation results show that compared with the traditional orthogonal iterative algorithm,this algorithm avoids the problems of iterative non convergence or poor convergence caused by data deterioration or improper selection of initial value,and improves the calculation efficiency by 4.6 times.The anti noise performance of the algorithm has been effectively improved,and the accuracy of this algorithm is higher under the same noise level,which shows that this algorithm is not sensitive to noises.The position error is better than 0.964 mm and the angle error is better than 0.765°.Taking into account the field measurement task,combined with the motion characteristics of rigid body,a multi-point cooperative target is designed,and an Euler-quaternion attitude solution model based on multi-point cooperative target is established.The model avoids the singular problem of Euler angle in attitude solution,and reduces the nonlinear error introduced by trigonometric function when Euler angle is directly used to calculate.Aiming at the problem of point target occlusion,a multi-target point matching algorithm with the same name based on the strong tracking UKF linear imaging is proposed.The simulation results show that the strong tracking UKF algorithm can achieve the accurate tracking ability of attitude information and complete the matching of image points with the same name when the state of the target changes suddenly or the mark points are occluded. |
PDF Full Text Request