Research On The Positioning Method Of Air-borne SAR Image Without GCPs

Air-borne SAR(Synthetic Aperture Radar) system is flexible and adaptable. The all-day, all-weather and high-resolution characteristics make it very useful in recent military applications by using it to take images of some target areas. In order to efficiently monitor the target areas, some important and interesting information must be extracted. And the precise position information is often needed in the SAR image. Thus the research of positioning method of the SAR image must be done. Moreover, the traditional positioning method is often based on the ground control points(GCPs), nevertheless in deserts, seas or the areas beyond the borders, the GCPs is very difficult to obtain. So the research of the positioning method of SAR image is really useful, important and worth to be carried out. Research on the positioning method of air-borne SAR image without GCPs is discussed in this thesis, and the following aspects are included:1. A roughly positioning method of air-borne SAR image based on imaging geometric model without GCPs is proposed. There are mainly two positioning methods without GCPs of air-borne SAR image: F.Leberl Model and R-D Model. Because the flight height of air-borne SAR platform is much smaller than earth radius, the curvature of the Earth can be ignored and it’s proper to choose the F.Leberl Model as the basic model. Then by combining the basic model and the imaging geometry of air-borne SAR, the roughly positioning method without GCPs is proposed with the data obtained by the combined system of Global Positioning System(GPS) and Inertial Navigation System(INS). Factors which influence the positioning accuracy of the method, especially the measured data of the combined system, are analyzed. And the numerical experiments’ results demonstrate the validity of the method and the analysis. As this method defines the geographical coordinate as the fundamental coordinate, it can be applied to the situation when the flying direction is not parallel with the horizontal plane. Besides, a judging method of the target’s position in this method is proposed and the iterative algorithm can thus be avoided.2. A precise positioning method of air-borne SAR image based on the registration of the previous roughly positioned SAR image and the corresponding optical image is proposed. As optical images are easy to be comprehended and well positioned, we can firstly find the cross points of lines of the roughly positioned SAR image and the corresponding optical image. Then matching points of the two images can be extracted from the cross points with the method of Random Sampling Consensus(RANSAC). As a result, we can get a precise positioned SAR image. However, it still has some shift and rotation. Considering these characteristics, the registration of SAR image with optical image of mutual information is analyzed. In addition, by using the mutual information method, the rough SAR image is registered with the corresponding optical image, which is well positioned. So the precise positioned SAR image is gotten.