Extension of D-TomoSAR for multidimensional reconstruction based on polynomial phase signal

Multi-dimensional reconstruction of ground is one of the most important applications of differential synthetic aperture radar tomography (D-TomoSAR) system. The ground deformation model is considered to be linear in classical D-TomoSAR system model. However, the real deformation in practice is deviated from the linear model, and the deformation parameters cannot be estimated accurately by spectral estimation. Here, a polynomial deformation model is proposed to approach the real motion of the ground. Since the deformation history appears as a phase term in the model of D-TomoSAR system, the corresponding model could be extended to a two-dimensional (2D) polynomial phase signal (PPS) by introducing the polynomial deformation model. In addition, an effective method combining the 2D product high-order ambiguity function with 2DRELAXation (RELAX) algorithm is provided to estimate the coefficients of this 2D-PPS. Theoretical analyses and experimental results show that the ground elevation, linear and non-linear motions can be both estimated accurately, and it is possible to separate scatterers distributed along the slant height within a resolution cell with the proposed method.