Automatic detection of earthquake damaged areas from aerial images and digital map
Hiroaki Simizu, Toshiyuki Gotoh, Hitoshi Saji · 2004
When a natural disaster, such as a big earthquake, happens, we should grasp the range and seriousness of the damaged areas for organizing the rescue operation. The aerial images are useful to grasp the damaged areas widely and rapidly. In this study, we propose a new method of automatic detection of areas damaged by an earthquake from aerial images. In this method, we use two aerial images that are taken before and after the earthquake and we use a digital map of the area. First we register each aerial image with the digital map. Then we compute the differences of registered pixels of two aerial images, and we extract the damaged areas having large color differences on registered pixels. The damaged areas are detected automatically and correctly from the aerial images by our proposed method.