Abstract: Aiming at overcoming the disadvantages of the limitation of class number and misclassification while classifying terrain and land directly by using four components decomposed by Yamaguchi target decomposition method from PolSAR images，a novel unsupervised classification algorithm for full polarimetric SAR images based on four-components scattering model and combined FCM theory is proposed and is applied to deal with L-band PiSAR image.This algorithm uses the four components as input features of FCM cluster.The experimental result demonstrates the accuracy and effectiveness of this algorithm.