3-D SAR Imaging With Missing Data Along Height Direction

Three-dimensional Synthetic Aperture Radar（3D-SAR）can form a two-dimensional virtual large aperture in azimuth and height dimensions through the two-dimensional plain sweep motion of the sensor platform,and combine with the large bandwidth signals transmitted from the range dimension,so as to have three-dimensional resolution ability.The realization of three-dimensional imaging of the target has important application value in terrain mapping,safety inspection and other fields.However,in order to improve the efficiency of data admission and reduce the complexity of the hardware system,the sparse plain scan mode is usually used in the high dimension,resulting in serious missing of the echo height direction data,which brings great challenges to 3-D imaging processing.Three-dimensional SAR imaging with missing data along height direction,which refers achieve high-precision three-dimensional imaging when the radar scanning sparse trajectory in the high dimensional data missing problem,utilizing the prior characteristics of echo to realize echo reconstruction.This article focuses on the research of echo modeling,echo characteristic analysis,and 3-D imaging algorithms involved in 3-D imaging of SAR with missing altitude data.The main research contents are as follows:1.The geometric and echo models of SAR with missing altitude data are constructed,and the impact of missing altitude data on 3-D imaging is analyzed from two aspects:the degree of missing altitude data and the way of missing altitude data.This supports the subsequent research on imaging algorithms and illustrates the necessity of this study.2.A low-complexity 3-D imaging method based on low-rank prior is proposed.The complexity of three-dimensional echo data recovery is reduced by performing distance screening on the echoes after pulse pressure.Utilizing the low-rank characteristics of the echo height-azimuth section,the improved matrix completion is performed on the section with echo information,which solves the problem that the matrix completion cannot be directly applied to the lack of height-dimensional data,and realizes low-complexity 3-D imaging when the data missing degree is low.3.A high-precision 3-D imaging method based on sparse-low-rank prior is proposed,and the sparse-low-rank characteristics of the echo height-distance section are deduced.The truncated Schatten-p norm and l₀ norm are used to constrain the echoes with low rank and sparseness respectively,which avoids the single constraint caused by insufficient prior information.Then,the Hankel matrix structure is introduced to solve the problem that it is difficult to achieve high-precision data recovery when the data missing degree is high.4.Due to the problem that it is difficult to initialize the parameters reasonably in the TSPN-l₀ algorithm,a parameter optimization method based on TSPN-l₀ unrolling network is studied.The learnable parameters of matrix recovery algorithm TSPN-l₀ are determined by analysis,and the iterative optimization process of the algorithm is transformed into the structure of neural network,which provides a guarantee for the accuracy of matrix recovery.The above models and methods have been validated by simulation data and measured data.The experimental results show that the method proposed in this thesis can effectively achieve high-precision three-dimensional imaging of SAR with high dimensional data loss.