After a large-scale earthquake disaster, we must be able to determine the routes of vehicles for rescue operations. For this purpose, we must rapidly collect road information at an early stage over a wide area. To achieve this purpose, aerial images will be useful, because they enable us to rapidly collect information over a wide area and are unaffected by the disaster. We propose a method of extracting road blockage areas in an urban area using aerial images obtained after an earthquake disaster and a digital map. In this method, using edge and color information, we extract damaged areas of roads and areas of collapsed buildings. Then, we correctly extract road blockage areas using the positional relations between the damaged areas of roads and the areas of collapsed buildings. We developed software on the computer, based on the proposed method, and we analyzed actual aerial images and evaluated the effectiveness of our method.
View full abstract