Considering that the slow convergence speed and even false convergence would appear in the conventional implicit space mapping (ISM) algorithm during the parameter extraction, an improved implicit space leapfrog algorithm is used to extract the implicit space mapping parameters, and the non convergence of the algorithm during the parameter extraction can be improved effectively. By improving the mapping relationship between the accurate model and the coarse model, the iteration numbers can be reduced significantly and be used to optimize the design of the micro-strip dual frequency band filter operating at WLAN. The center frequencies of the proposed filter are 2.45 GHz and 5.25 GHz respectively. The design requirements have been met after three iterations, thus the feasibility and effectiveness of the proposed algorithm has been verified.