This paper introduces an indoor global localization method by extending and matching features. In the proposed method, the environment is partitioned into convex subdivisions.Local extended maps of the subdivisions are then built by extending features to constitute the local extended map set. While the robot is moving in the environment, the local extended map of the current local environment is established and then matched with the local extended map set. Therefore, global localization in an indoor environment can be achieved by integrating the position and orientation matching rates. Both theoretical analysis and comparison experimental result are provided to verify the effectiveness of the proposed method for global localization. This paper introduces an indoor global localization method by extending and matching features. In the proposed method, the environment is partitioned into convex subdivisions. Local extended maps of the subdivisions are then built by extending features to constitute the local extended map set. While the robot is moving in the environment, the local extended map of the current local environment is established and then matched with the local localized map set. Therefore, global localization in an indoor environment can be achieved by integrating the position and orientation matching rates. Both theoretical analysis and comparison experimental result are provided to verify the effectiveness of the proposed method for global localization.