Abstract-Although MUSIC (MUltiple SIgnal Classification)-type algorithm has shown feasibilities as a non-iterative imaging technique of thin penetrable electromagnetic inclusion from its far-field multistatic response (MSR) matrix, it induces a poor result whenever one tries to obtain such inclusion of both dielectric and magnetic contrast with respect to the embedding homogeneous space R 2 case. In this paper, we develop an improved non-iterative imaging algorithm based on the modeling of multi-frequency MSR matrix according to a rigorous asymptotic expansion of the scattering amplitude. Numerical examples exhibit that presented algorithm performs satisfactorily for single and multiple thin inclusions, even with a fair amount of random noise.