Content-based retrieval for remotely sensed images

The work describes an innovative technique to automatically extract and manage remote sensing image-content. Simple but very flexible numeric recognition methodologies allow the content-based retrieval from huge remotely sensed image database. The most important result of this methodology is a tool for the information retrieval based on example. In order to properly characterize remotely sensed images and improve retrieval performance, many factors, such as the image resolution, the scale, the sensor features, have been taken into account. Kingfisher is the content-based database management system, developed at DIBE laboratories, that exploits these methodologies.