A robust and high-resolution two-dimensional (2-D) imaging algorithm is proposed for retrieving the shape of conductive objects embedded in a uniform lossy and dispersive medium, The target is modeled in terms of the location of points that represents the outer contour of the object, together with the parameters of the medium. A nonlinear least-squares fitting is applied to the estimated scattered waveform to adjust the model parameters. The estimated waveform is computed using the extended rag-tracing method, which incorporates the edge-diffraction waves. The performance of the algorithm is examined with numerical simulations and test-site experiments. The simulation with clutters also revealed the robustness of the algorithm even under fairly strong clutter environments.