The Gauss-Newton inversion (GNI), an iterative algorithm, is developed from the frequency domain to the time domain in order to simultaneously reconstruct the electrical permittivity and electric conductivity of a two-dimensional object of interest by directly using the ultra-wideband time-domain measurement data. The resulting forward problem is solved by the finite difference time domain method, while the ill-posedness of the corresponding inverse problem is restrained by an adaptive regularization technique at each iteration. Furthermore, the modified GNI algorithm is applied to four types of numerical examples where a noise model is considered, and the simulated results preliminarily demonstrate its feasibility and robustness. The reconstructed images present super resolution, thus it is expected to be used in the engineering practice such as the detection of the early-stage breast cancer.