This paper uses the ceiling vision and the odometry to achieve localization for mobile robots in indoor environment. Given that the fluorescent lights with rectangular lampshades are distributed regularly on the ceiling in most indoor environments, the corners of lampshades are selected as features to navigate the mobile robot. An efficient extraction algorithm is proposed to identify the corner features. In addition, the extrinsic parameters of the visual system are calibrated by our developed method. The approach to estimate the initial pose of the mobile robot is proposed, which is also used to provide the ground truth for the robot pose. Experimental results are included to verify the effectiveness and robustness of the proposed localization method.