Target 3D Reconstruction Technology Based On Spaceborne SAR Multi-angle Imaging

With the increasing demand for information of remote sensing targets,the accurate reconstruction of 3D information of scene to be observed has become a hot research direction in remote sensing field.The synthetic aperture radar(SAR)imaging technology not only loses the 3D information of the target,but also makes the 2D position of the target shift.At present,multi-orbit observation of spaceborne SAR with a single antenna can be used for 3D reconstruction of the observation scene,but this method has some problems such as long revisit period and complicated orbit control.In this thesis,the 3D reconstruction technology based on spaceborne SAR multi-angle imaging is studied,and the 3D information of target can be reconstructed exactly in a single orbit.The main work of this thesis can be summarized as follows.1.The geometric model of spaceborne SAR multi-angle observation and the projection process of the 3D space point target to the imaging plane are described in a clear and concise matrix form.The consistency of the range migration between a point target in 3D space and its two-dimensional image point target is obtained by using the method of geometric proof,the applicability of SAR imaging algorithm to point target in space and the necessity of non-linear track for multi-angle 3D reconstruction are confirmed.The back projection(BP)algorithm is used for SAR imaging,which verifies the applicability of BP algorithm to multi-angle SAR imaging and the correctness of equivalent observation center.2.A novel spaceborne SAR image registration method is proposed to deal with the sparsity of SAR images of point-structured targets.Starting from the sinc function and as an auxiliary to analyze the characteristics of the contourmap of the point target's impulse response and the selection of the amplitude parameter,a method of image point extraction based on the target contour is proposed;and this thesis analyzes the ability of the equivalent azimuth angle as a parameter to help the matching of the image points in two images,and proposes a matching method based on coordinate rotation transformation.The processing results of spaceborne SAR images show that the precision of image point extraction is at sub-meter level,and the matching method is correct and effective.3.A new formation model equation is proposed based on range projection geometry.Its form is simpler,the required parameters are less,and the geometric meaning is more obvious.Further more,the 3D reconstruction method in this paper is constructed by analyzing the equations.The simulation results of 3D reconstruction based on SAR imaging show that the reconstruction accuracy of this method is at leastmeters,and is close to 1 m under the condition of high SNR or high resolution.At the same time,the effects of observation angle,orbit altitude,image point error and resolution on the reconstruction accuracy are simulated and analyzed,and the applicability of the reconstruction method in the traditional stereo image pairs technique is verified.