Abstract: The apparent difference of gray value between impulse noised pixels and the pixels around them causes the neurons of intersecting cortical model （ICM） fired at different stage.In this paper we used the pulse output generated from similar neurons synchronized fire mechanism in the ICM to locate the impulse noise inside the image,and adaptively filtered the impulse noised pixels.The signal to noise ratio （SNR） generated from the ICM filter proposed in this paper is increased by50％ to60％ than those from other nonliner filters,and the computation time is only about 20％ to30％ of them.In the meantime,the proposed algorithm can effectively preserve the details of image,which is greatly helpful for further image processing and recognition.