Research On Satellite Image Positioning Without Ground Conrol Point Assisted By DEM

Traditional remote sensing images positioning requires precise ground control points,ofwitch the number and distribution are also impotant.But for some region such as abroadterritory,offshore islands (reefs),forest,desert,etc.,where the operation personnel are hard to reachor the ground features are not obvious, the acquisition of ground control points or point prickingare often very difficult or even impossible to complete.On the other hand, as acquisition of thecontrol points takes a long period, using the control points for image positioning can not meet theneeds of counter-terrorism, disaster relief and other emergency mapping support tasks. Therefore,the study of satellite remote sensing images positioning without ground control points hasbecome an essential and urgent task.In this paper, on the basis of the analysis of high-resolution satellite remote sensing imagepositioning without ground control points, designed a program that using DEM data assistsremote sensing images positioning without ground control poinst through terrain matching, andhave done a more in-depth study and discussion about terrain matching, stereo image positioningand systematic error compensation. The main content and innovation:1. Sdudy on the rational function model (RPC)-based stereotactic positioning methodwithout ground control points, the case of, proved by experiments in the absence of the controlpoint the (RPC)-based stereotactic positioning has a large systematic error. Global public SRTMand GDEM data were analyzed, and their accuracy were evaluated.2. Satellite images positioning Without ground control point generally poor in absoluteaccuracy, and better in relative accuracy, to design a DEM-asisted remote sensing imagepositioning method, match the DSM obtained without ground control points with DEM, such asSRTM, GDEM, other existing DEM data in survey area, set up error compensation model tocompensate the systematic error.The experimental results show that this method can greatlyimprove remote sensing image positioning accuracy without ground control points, and reducedependence on ground control points.3. Designed a SIFT matching and least squares matching (LSM)-based multi-scale DEMmatching method, the method will convert DEM to grayscale images. As SIFT hasscale-invariant characteristics, use SIFT mathch to get a crude matching result. Finally carry onleast squares match for precise matching to improve the DEM matching accuracy.