Abstract
Conventional photo retrieval approaches were inefficient to find significant photos in ground-based monitoring using a large amount of photos taken after The Great East Japan Earthquake. Therefore, we focused on a new photo retrieval methodology improve the efficiency in a disaster monitoring. Our methodology is consisted of a photo geocoding using GPS and magnetic sensor data, geospatial arrangement of geocoded photos, and georeference rectification based on Scale-Invariant Feature Transform. Moreover, we have confirmed that our methodology can facilitate control and browsing of geocoded photos.
