Robotic mapping and exploration is basic to many robotic applications such as search and rescue operations in disaster scenarios. Multi-robot systems can speed up exploration tasks in such critical situations by making use of distributed sensors to increase the range of exploration and mapping. In this case, every robot explores and maps different areas that are finally merged and connected. To build a map of an unknown environment, a robot must perform SLAM or Simultaneous Localization and Mapping and based on the perceived data metric maps such as feature maps or occupancy grid maps are constructed. The maps are then used for localizing the robots and path planning. Although metric maps allow precise robot localization and estimation, they suffer from high memory requirements needed to store all the information in occupied cells. Moreover, merging maps from other robots is an intensive process. On the other hand, topological features map representation can be used to store information into nodes and edges and does not have any large memory requirements. In this paper, we present a combined metric-topological mapping approach to multi-robot SLAM. This method maintains a graph with sensor information stored in nodes and edges that can be optimized globally. By combining local metric and topological maps build by individual robots, the graph structure can be merged and extended to map large areas effectively. To robustly merge local maps into a global one, we used visual features from each robot that are matched in a distributed system. The graph node-edge structure is used for path planning and information sharing between robots resulting in optimized task distribution.