Study On Some Key Techniques Of Satellite Photogrammetry Without Ground Control Points

China has a vast territory and a large number of no man's lands such as the plateau and the desert.In addition,with the expansion of our country's overseas interests,global photogrammetry without ground control points is our first choice.The Mapping Satellite-1 is China's first-generation transmission stereoscopic mapping satellite,which photogrammetry target is to achieve the accuracy requirement of measuring 1:50000topographic maps without ground control points.The paper starting from the principle of photogrammetry of the three-line CCD image of the Mapping Satellite-1,theoretically discusses and experimental studies the key techniques of the photogrammetry parameter calibration,low-frequency error compensation and beam method large-area block adjustment under the condition of no ground control point.The main research work of the paper is as follows:1.On the basis of studying the imaging and photogrammetry principles of the three-line CCD camera of the Mapping Satellite-1,the mathematical model of three-line array photogrammetry was established.Factors affecting the accuracy of image positioning were analyzed,and the Star-Ground link error simulation and evaluation model was established.The paper theoretically evaluates the impact of various errors on the positioning accuracy from the physical model level,and uses the test field data of the three-line array image of the Mapping Satellite-1 to perform quantitative error analysis by the key factors affecting the image stereo positioning accuracy,evaluates the effects of various errors on the positioning accuracy.The results of simulation verification and theoretical analysis were consistent by comparing with them.2.Based on the strict mathematic model of three-line CCD image photogrammetry,the geometric calibration based on the variation angle of the Star-Ground camera and the geometric calibration of the change values of the internal parameters in the three-line array camera were studied.The variation of the angle was calibrated as the bias matrix.The variation of the internal orientation element was calibrated by a constant,and the uncontrolled positioning accuracy is greatly improved;Accurately taking into account the aberration shift,the effect of the internal parameter calibration based on different additional parameter models on the positioning accuracy is explored,and the verification is made which the distortion of the internal parameters of the Mapping Satellite-1 is mainly reflected in the change of the principal point and the principal distance of the optical lens,and the distortion associated with the linear CCD is negligible;a joint on-orbit geometry calibration technology based on multiple strip images is proposed,and multiple strips are simultaneously which can weaken the effect of low-frequency error on external detection,making the Star-Ground camera angle more close to the real value;Self-calibration block adjustment based on the common model is proposed to solve the problem of uncontrolled positioning accuracy improvement of the Mapping Satellite-1 three-line CCD image 1A-level RPC product without attitude and or bit data.3.The causes and manifestations of low-frequency errors were analyzed,and the effects of low-frequency errors on positioning accuracy were also described.The principle of low-frequency error compensation for angle exterior elements the corner was studied,the general correction model and special solution for low-frequency errors were proposed.The low-frequency calibration can basically eliminate the plane and elevation error of the ground point and complete the “last kilometer” key problem of the uncontrolled positioning accuracy improvement;at the same time,combing calibration process of the traditional calibration and dynamic calibration,and using the Star-Ground camera angle without calibration,it is directly verified by the extension method of dynamic calibration of the angle exterior orientation elements.Finally,the errors of the low-frequency measurement error,the tilt error of the three-line image plane and the variation of the Star-Ground camera angle were discussed,and the errors of the traditional periodic calibration were corrected and redefined.4.The method of strip block adjustment based on the strict geometric model is studied to eliminate the star random error in the strips and form 1B-level image product.Because of the roll angle cannot be eliminated within the strip by the low-frequency measurement error,the systematic errors and random errors between the strips can be eliminated by a large-area block adjustment method based on the rational function.The error within the entire tends to be uniform,and the positioning accuracy is the highest.For hyperscale large-area models,the nature of the sparse matrix is fully exploited,the normal equations are modified by using the feature of real symmetric matrices which the block adjustment equation was,and the matrix operations are optimized concurrently in the solution process to improve computational efficiency by reducing memory.