| In the past decade,with the implementation of a series of major special projects such as “Earth Observation with High Resolution” and “National Civil Space Infrastructure”,satellite remote sensing technology has developed rapidly in China.We has successively launched dozens of optical remote sensing satellites of various series,such as “Zi Yuan”,“Gao Fen” and “Tian Hui”,providing rich data resources for global mapping,and has strongly supported the scientific development of China's economic construction,environmental protection,disaster warning,defense and military,and many other fields.With the increasing demand for global high-precision mapping,the geometric precision processing technology of optical satellite remote sensing image(OSRSI)based on system error calibration and block adjustment(BA)is becoming more and more important.However,most areas of the earth's land surface are not only lack of high-precision geographic reference data,but also are no man's land or overseas areas where ground control points(GCPs)cannot be obtained.As a result,the traditional geometric precision processing method relying on evenly distributed GCPs is no longer applicable.When lacking the control of GCPs,the system error calibration based on image resection is difficult to obtain accurate results due to the lack of enough observations,and the BA system of OSRSIs is short of absolute reference datum,resulting in the weak robustness of the whole adjustment system.Moreover,the OSRSI has the characteristics of long focal length and narrow field of view,which makes the systematic geometric error of OSRSI cannot be compensated by additional self-calibration model,or suppressed in BA,resulting in the accumulation of errors,and the deformation of the block.In addition,due to the lack of absolute constraints and a large number of weak intersections in the BA system,the adjustment function is deficient in rank.Besides,there may be images with abnormal observation conditions in the complex block.The problems that do not appear under the constraint of GCPs will seriously affect the BA under the condition of lack of GCPs,leading to the decrease of BA accuracy and reliability.This paper conducts the research on the quality control method of geometric precision processing for OSRSI under the lack of GCPs,the main research contents and innovations are as follows:1)This paper analyzes the error sources and characteristics in the geometric precision processing of OSRSI,establishes the accumulation model of the systematic geometric error in BA,and analyzes its cumulative effect,which provide the theoretical basis and model basis for the follow-up calibration of systematic errors and quality control of BA.According to the imaging principle of linear push-broom optical satellite,the physical geometric imaging model and rational function model are established to lay the foundation for the subsequent analysis and processing.Based on the physical geometric imaging model,the errors and their characteristics are analyzed,and the influence of various errors on the geometric quality of single scene image and the quality of BA system is further analyzed.Further,this paper establishes the error accumulation model in BA,and analyzes the cumulative effect of systematic errors.The validity of the above analysis and models are verified through a group of experiments,which lays the research direction and provides theoretical basis for adopting pre-calibration to solve the cumulative effect of geometric errors in the BA.2)This paper proposes a geometric calibration method for the systematic errors of OSRSI based on self-constraint between images,which achieves the high-precision compensation of systematic errors under the condition of lacking GCPs,and guarantee the internal geometric accuracy of input images of BA.Aiming to the error accumulation caused by systematic internal geometric errors of single image in BA lacking GCPs,a scalibration method for systematic error in OSRSI is proposed based on the selfconstraint between images.Based on the established unified calibration model considering side view imaging,the overall iterative calibration for reference image,the overall calibration based on bundle adjustment for non-reference image,and overall calibration strategy for stereo image pairs are proposed.Therefore,the systematic errors of OSRSIs are compensated accurately without using the reference data of ground calibration site,further the internal geometric accuracy of the input images in BA can be guaranteed.3)A quality control method of the BA for OSRSIs under the condition of lacking GCPs is proposed to ensure the accuracy and reliability of BA.Aiming at the quality control of the BA lacking GCPs,a BA model based on virtual weighted observations is proposed,and on this basis,a weight determination strategy of observations considering the weak intersection of tie points is designed to ensure the robustness of BA optimization.Secondly,an automatic method for detecting the image of abnormal observations based on variance estimation is introduced,to guide the optimization of internal connection conditions in BA.Finally,a BA optimization strategy based on elevation error compensation is proposed,which can correct the elevation error and overcome its cumulative effect,ensuring the accuracy of BA.In view of the above research content,this paper uses three linear array images of ZY-3 satellite and backward images of Gaofen-7 satellite to verify the self-calibration method,and uses two linear array images of Gaofen-7 satellite to verify the internal quality control method of BA.The experimental results show that the proposed self-calibration method can effectively compensate the systematic geometric errors and improve the internal geometric accuracy of satellite images.Furthermore,the internal quality control method proposed can ensure the reliability and accuracy of BA lacking GCPs,and achieve high-precision geometric processing of optical satellite images. |
PDF Full Text Request