| The popularity of Photogrammetry System for micro UAV greatly widens the applications of photogrammetry.But the orientation stability of this kind of UAV is poor.In order to ensure the accuracy,a large number of ground control points are usually needed in the process of UAV aerial survey,which has become the most stressful jobs in the whole aerial survey process.How to minimize the dependence on the ground control points has been a hot spot in the field of photogrammetry,and the accurate exterior orientation elements is the key to the realization of sparse ground control and even no control.However,due to the limitation of load and device cost of UAVs,the accuracy of exterior orientation elements obtained by micro UAV is poor,which can not meet the requirements.In order to solve this problem,the paper focuses on the optimization method of the exterior orientation elements using the differential GNSS and the structure from motion(SFM),and the experimental verification is carried out,and an effective approach to the small range photogrammetry using micro UAV with few ground control points is explored.The main work and innovation are as follows:1.In view of the low accuracy of the exterior orientation elements in the micro UAV images,a LAMBDA ambiguity resolution algorithm based on the BDS /GPS is proposed,and a three curve interpolation algorithm for recording the location of the camera with accurate exposure is proposed in combination with the method of the phase center eccentricity measurement of the analytic geometry GNSS antenna.The experiment shows that the method can greatly improve the ratio of the post differential fixed solution,and can eliminate most of the camera delay errors,and finally obtain the exterior orientation elements with centimeter-level accuracy.2.A robust Lie algebraic rotation averaging global SFM algorithm(G-SFM)is proposed to optimize the exterior orientation elements of UAV images.First,the image overlap is calculated by using the high accuracy camera station coordinate GNSS information,the image relation map is constructed and the SIFT features are matched,then the image orientation parameters are iteratively solved.By analyzing the distortion error of the digital camera,the orientation error and the other system errors,the self-calibration bundle adjustment model is constructed with high accuracy GNSS camera station constraints which is used to solve the optimal value of exterior orientation elements and camera parameters under the optimal GNSS weight.3.In order to verify the effectiveness of the aforementioned UAV image exterior orientation elements optimization method,experiments are carried out on micro UAV images with different terrain and different measuring scales.In the 1:1000 experiments in plain and mountainous areas,the results of aerial triangulation with the raw position and orientation data,traditional differential GNSS data,and the G-SFM optimized exterior orientation elements data were compared.The experiment shows that after the optimization of G-SFM,4 control points can meet the aerial triangulation specification well,and the dependence on the number of control points is obviously reduced.In the experiment of Songshan UAV calibration field,the results of three different types of UAV data processing are compared,and find that without the use of ground control points to participate in the adjustment calculation,the G-SFM algorithm has the best control effect to the elevation error he use of control points in the adjustment.However,this method has the best effect on the error control of the checkpoint elevation.Compared with the row position and attitude data,the differential GNSS data supported aerial triangulation,the number of checkpoints in elevation error can be reduced by 82% and 39%,greatly improved the elevation accuracy of UAV image aerial triangulation without image control points. |
PDF Full Text Request