Inverse scattering of inhomogeneous dielectric objects

• Main Authors: Rong-Hwa Yang, 楊榮華
• Other Authors: Chien-Ching Chiu
• Format: Others
• Language: zh-TW
• Published: 1995
• Online Access: http://ndltd.ncl.edu.tw/handle/09046787071234519929

碩士 === 淡江大學 === 資訊工程研究所 === 83 === The inverse scattering of inhomogeneous dielectric cylinders is investigated. Based on the unrelated illumination method, two inverse algorithms for reconstructing the permittivities of a biaxial inhomogeneous dielectric object and an isotropic inhomogeneous dielectric object which includes given shape conductor are proposed in this thesis respectively. The inversion algorithms are based on the rigorous mathematics. Assume that a group of unrelated incident waves produced by beam focusing scheme is incident upon the cylindrical object of unknown permittivities and the scattered field is measured outside. Based on appropriate boundary condition a set of nonlinear integral equations is derived. The moment method is then used to transform the integral equations into matrix form with pulse basis functions for expanding and Dirac delta functions for testing.The key problem for the inversion algorithm is that the incident matrix must not be a singular one, in other words, all the incident column vectors that form the incident matrix must be linear unrelated. As a result, the dielectric constant distribution of the object can be reconstructed when the object is illuminated by a group of unrelated incident wave. Two-dimensional objects with three different cross-sections which include rectangular, square and circular shape are considered in the numerical simulation. The integral equations are solved by moment method to generate the scattered electric field. Note that the discretization number for the direct problem is four times as that for the inverse problem. The real part and imaginary part of the noise with random uniform distribution is added to the complex scattered field. Numerical results demonstrate a very good noise- rejecting capability of this scheme. Good reconstruction has been obtained even in the presence of additive random noise in measured data. In addition, the effects of noise on the reconstruction result are also investigated.