Direct Georeferencing Of Airborne Sensors In National Frame

Since GPS/INS integrated navigation was successfully introduced to the field of photogrammetry in the early1990s, the data processing flow of georeferencing has undergone a thorough change. After carefully calibrating the geometric relationship between an imaging sensor, a GPS receiver and an INS, airborne position and orientation system (POS) can accurately output the exterior orientation parameters (EOPs) of the imaging sensor, which can be directly used for topographic mapping. This method is often called direct georeferencing (DG). Unlike the traditional georeferencing approach (i.e., indirect georeferencing), ground control points (GCPs) and aerial triangulation (AT) are no longer needed in DG, and both office and field work can therefore be greatly reduced. However, due to the lack of constraints by GCPs and tie points (TPs), the ground coordinates calculated by DG are very susceptible to various systematic errors.Aerial photogrammetry data products are typically required in national coordinates, and therefore the national frame is usually employed to conduct the data processing of georeferencing. However, as a national frame is not Cartesian but rather adopts a conformal map projection, the georeferencing process is inevitably affected by many geometric distortions. This issue is particularly serious for direct georeferencing because it has the nature of extrapolation and various geometric distortions will directly propagate into ground coordinates. Therefore, all major geometric distortions must be carefully corrected to ensure the accuracy of DG result in national coordinates.This paper focuses on the problem of direct georeferencing of airborne sensors in national coordinates, which mainly includes:1) Theoretical analysis and accuracy assessment on different methods of transforming airborne POS data to a national frame.In order to perform DG in national coordinates, the first step is to accurately transform the trajectory outputted by airborne POS data processing software to the required national frame. This paper derives all formulas of three representative attitude transformation methods (i.e., rotation matrix method, compensation matrix method, and coordinate transformation method). A high-precision experiment is specifically designed for assessing the accuracies of these attitude transformation algorithms and two commercial software (POSPac from Applanix and IPAS CO from Leica). Experimental results show that the transformation errors of the rotation matrix method are completely negligible and its computational cost is usually lower than all other algorithms, and therefore it is the most recommended method to transform airborne POS attitude data to a national frame. On the other hand, the performance of the two commercial software is not very satisfactory because the maximum error in their transformation results already exceeds the theoretical measurement error of airborne POS.2) Geometric distortion model of direct georeferencing in national coordinates.In this paper, all the major seven DG distortion factors in the national coordinates are identified and grouped into four categories, including one3D scale distortion (the datum scale factor distortion), one height distortion (the earth curvature distortion), two length distortions (the horizontal-to-geodesic length distortion and the geodesic-to-projected length distortion), and three angle distortions (the skew-normal distortion, the normal-section-to-geodesic distortion, and the arc-to-chord distortion), where the datum scale factor distortion, the skew-normal distortion, and the normal-section-to-geodesic distortion have not been reported in the literature.3) Geometric distortion correction algorithm for the direct georeferencing of airborne LiDAR (Light Detection And Ranging) data in national coordinates.This paper derives high-precision and practical map projection correction formulas for the DG process of airborne LiDAR data in national coordinates. Experimental results prove that the accuracy of the practical correction formulas is much better than the traditional formulas, and only25%more computational cost is required. On the other hand, the residuals of the high-precision correction approach are practically negligible.4) Geometric distortion correction algorithm for the direct georeferencing of aeiral imageries in national coordinates.This paper derives the formulas of several traditional geometric distortion correction methods (including the traditional earth curvature correction method and two length correction methods:the changing flight height method and the changing focal length method) and presents several improved methods (including the equivalent vertical image method, the changing image coordinate method, the changing object coordinate method, and the forward intersection prediction method) for the direct georeferencing of aeiral imageries in national coordinates. Experimental results prove that the performance of the equivalent vertical image method is always better than the traditional earth curvature correction method because it is completely unaffected by image tilt and no extra computational cost is incurred. The correction results of the changing object coordinate method is not appreciably influenced by terrain undulations, and its accuracy is usually better than the two conventional length correction methods (i.e., the changing flight height method and the changing focal length method). The correction residuals of the forward intersection prediction method are always trivial, and it can therefore meet the needs of high-precision applications.