Extract spatial information from High-Resolution remote sensing images is the focus academicresearch of remote sensing and digital photogrammetry.Using the satellite remote sensing images,wecan high-precision target the three-dimensional surface date of the earth.In order to achieve this,it isessential to establish the sensor imaging geometry model,understand the mathematical relationshipbetween the earth coordinate and image plane coordinate.The measurement error in this phase willgreatly affect a series of follow-up analysis and decision.Therefore,it is the most important technologyto establish the relationship model between the remote sensing image and the surface of earth.Currently,it is always use the linear array CCD sersor to make a high-resolution remote sensingimage.The push-broom imaging mode makes the scan lines of the exterior orientation changeaccording to time function,and its narrow beam and small base height ratio make strongly correlationbetween the oriention parameters,making the least squares estimates have larger variance is no longerthe best estimates.In order to establish the geometry model of the linear CCD image sensor, we mustovercome the strong correlation of orientation parametersIn this paper,we pose the dual quaternion which is a new mathematical tool to finish thegeometric positioning of linear CCD images.Build the strict imaging geometic model,use the dualquaternion to depict the spiral movement in space,then we achieve the transformation from imagepoint coordinate to ground point coordinate.Using this method,we can overcome the strongcorrelation of the orientation parameters and improve the geometric positioning accuracy.In order toensure the new theory,we make a lot of experiments based on a number of High-resulotion linearCCD images.Result shows that it can be used to restore the three-dimensional information of the earthusing the remote sensing images,and it also can be used to positioning the global goals and extractingthe basic geographic datas. |