Research On Combined Positioning Model Of Multi-source Heterogeneous Images

Based on the summary of the positioning model of sensor images of different carriers andimaging mechanism, this dissertation reconstructed the imaging equation of aerial/spaceoptical/radar images with unitive external orientation parameters in ECR coordinate system. Onthe basis of the imaging equations, orbital model and measurement data processing, an integrated“Photographic triangulation” technique of multi-source and heterogeneous remote sensing imagewas studied to achieve the high precision of combined positioning of multi-source andheterogeneous remote sensing images with sparse (no) ground control points by configuringsensor parameters and adjustment parameters. The major tasks and contribution of thedissertation are:1. The Imaging Equation and the Positioning Model of Aerial Optical Image in the GeocentricCartesian Coordinate System was constructed.In this dissertation, the positioning model of aerial remote sensing image was constructed inthe geocentric cartesian coordinate system. The influence of curvature of the earth cannot betaken into account with respect to earth ellipsoid tangent plane cartesian coordinate system andmap projection coordinate system, which reduced the transformation of POS data, wasconducive to the combined adjustment of multi-source images and data. The method ofmulti-attitude reference datum was proposed, and the problem of singular and multiple solutionsin calculating the attitude angle were conquered.2. The Imaging Equation and the Positioning Model of Space Optical Image in the Earth-fixedGeocentric Cartesian Coordinate System was constructed.In this dissertation, the imaging equation of space image was directly built in the earth-fixedgeocentric cartesian coordinate system by adaptively processing of the attitude angle despite therotation of the earth and other parameters, making the imaging equation in the geocentriccoordinate system more simple and more adaptable; the orbital correction model was built in theflight coordinate system, adapted with laws of orbital system errors, which was more suitable forthe positioning of large-span space remote sensing image. In this study, the internal relationbetween the positioning model based on the collinearity equation and that based on the viewvector of image elements has been established, which makes the orbit and attitude parameters ofspace remote sensing image more Refined, and the direct and inverse solution of coordinates ofimage points and ground points much easier.3. The Imaging Equation and the Positioning Model of Radar Image was Constructed based onElements of Exterior Orientation. With elements of exterior orientation as the orientation parameters, in this study a newimaging equation of radar image was constructed－Range-Coplanarity (R-Cp)equation. Theimaging equation reflected the imaging mechanism of the radar images in the direction of rangeand the direction of orientation, and manifested the attribute of the image point coordinate as anindependent observed value of photogrammetry so that the positioning of radar image are able toeasily adopt the mature algorithms of photogrammetric data processing of optical image. On thebasis of R-Cp equation, this study provided a set of integrated positioning theory and method ofaerospace side-looking radar images including correction, three-dimensional positioning, areanetwork adjustment, and pre-processing of matching geometric.4. The Positioning Technology and Ways of Multi-source Remote Sensing Image was Realized.With elements of exterior orientation as the unified orientation parameters, the explicitfunction form of image point coordinates as unfied imaging equation, the geocentric cartesiancoordinate system as the unified data-processing frame, polynomial as the unified orbit andattitude corrective model and combined adjustment as the unified data-processing method, by thearrangement of sensor, in this study a photographic triangulation model suitable for differenttypes of sensors remote sensing image was built.The paper also put forward a method ofRigorous image Positioning for sensor without idiographic parameters based on equivalentsensor parameters,and several ways getting equivalent sensor parameters were introduced.5. Several Key Technologies of Calculation and Accuracy of the Positioning of Large-scaleRemote Sensing Images Were Researched.In this study, a direct filling method of the coefficient matrix of the equation of the changemethod without the unknown number of the ground point coordinates was provided, omitting thestorage space of the coefficients of the error equation and the coefficients of the unknownnumber of ground point coordinates in normal equation. The decorrelation method of orientationelements of image positioning was generalized and analyzed and some instructive conclusionsabout combined adjustment of the positioning of remote sensing image were drawn. Thesimulation technology of computer was introduced into the analysis of the factors that affect theerrors of three-dimensional positioning of remote sensing images and the analysis of influence ofthe correlation among orientation parameters, which provides a quantitative, visual analysis toolfor the factors influencing the positioning accuracy of the complicated remote sensing image.6. The test verification of the proposed theoretical methodOn the basis of the theoretical method proposed in this study, the software prototype withstrong versatility, which can be applied to the rigorous positioning of multi-sensor images, wasdeveloped. The positioning of the multi-sensor images was realized through the configuration ofsensor parameters. The singe and integrated positioning of the representative nine types of sensor images were tested, which was analyzed to draw some meaningful conclusions. The test alsoproved that the model built in the study was feasible.