Artificial immune systems and aerial image segmentation

Abstract

This paper describes our initial investigations in applying artificial immune systems to feature segmentation in remotely sensed images. The current generation of commercial imaging satellites provides increased opportunities for automated image analysis due to the large volume of high resolution imagery they will produce. Artificial immune systems (AIS) are successful in other pattern recognition tasks and have several similarities to the aerial image classification problem. We use an AIS for road pixel identification and observe several areas for further development.