Development of Power Distribution Facility Map Input System Using Automatic Image Recognition Technology

In this paper, we propose a recognition technique for electric poles and wires that are drawn on a power distribution facility map. The map represents the power distribution supply facilities in Korea Electric Power Corporation (KEPCO), and the proposed technique is to be used for automatic input to the huge geographic information system (GIS) database. This technique is based on circularity and connectivity of image recognition and consists of four progress phases. Initially, we extract a power distribution facility area from the input image using a threshold value. Secondly, we perform a thinning process. At the third step, we extract candidate poles based on circularity. Finally, electric wires between two branch points are recognized based on connectivity. Candidate poles with at least one wire are confirmed to be electric poles. The proposed technique has been tested on many power distribution facility maps and the recognition success rate is over 85%. We made an auto GIS database input system based on this technique and performed field tests successfully.