To informatively plan optimal paths for autonomous mobile robots in indoor environment is essential in real life cases. In view of the shortcomings of the traditional path planning strategies based on the cameras mounted on robot body and considering surveillance cameras are always ready and available sensors in indoor environment, a real-time navigation system based on vertical surveillance camera is proposed. Based on the assumption that the surveillance camera is able to capture the full picture of the environment, for each frame of the video recorded by the camera, a traversable node map of the environment for the robot is firstly established. Background subtraction is then applied, to achieve real-time localization of the robot in the node map, by comparing an empty environment with each frame of the video feed where the robot exists. Secondly, two path planning algorithms are applied on the localized node map, and a comparison is made for informative path planning. To evaluate the robustness of the proposed system, three experiments are conducted on a preset map, a differential caster wheel robot and a suite of matched motion planning system. The time for the robot to navigate itself from entrance to exit and number of nodes traversed for each experiment are recorded and compared. The final result demonstrates the robustness and efficiency of the proposed path planning method based on a vertical surveillance camera.