Global localization is a fundamental ability to recognize the accurate global position for a mobile robot in a revisited environment. The map-based global localization gives a precise position by calculating an accurate transformation, but the comparison with large 3D data is quite time-consuming. The appearance-based global localization which determines the global position by image retrieval techniques with similar structures is real-time. However, this technique needs external illumination constraint and does not work in the dark extremely. This paper proposes a combination of the map-based global localization and the appearance-based global localization. Instead of camera images used for the appearance-based global localization, we utilize reflectance images which are taken as a byproduct of range sensing by a laser range finder. The proposed method not only detects previously visited scenes but also estimate relative poses precisely. The effectiveness of the proposed technique is demonstrated through experiments in real environments.