Precise and robust localization is a significant task for autonomous vehicles in complex scenarios. In this paper, a novel method is proposed to precisely locate the autonomous vehicle using a 3D-LIDAR sensor. The curb-based feature matching and intensity-based feature matching results are fused to obtain an accurate estimated position. A curb detection method is proposed to extract the curb position and an area probability searching method is proposed to match the intensity feature. Experimental results demonstrate the accuracy and robustness of the proposed method.