Knowledge-Based Concepts for the Fusion of Multisensor and Multitemporal Aerial Images

Abstract

The increasing amount of remotely sensed imagery from multiple platforms requires efficient analysis techniques. The leading idea of the presented work is to automate the interpretation of multisensor and multitemporal remote sensing images by the use of common prior knowledge about landscape scenes. In addition the system can use specific map knowledge of a GIS, information about sensor projections and temporal changes of scene objects. Prior expert knowledge about the scene content is represented explicitly by a semantic net. A common concept has been developed to distinguish between the semantics of objects and their visual appearance in the different sensors considering the physical principle of the sensor and the material and surface properties of the objects. A flexible control system is used for the automated analysis, which employs mixtures of bottom up and top down strategies for image analysis dependent on the respective state of interpretation. The control strategy employs rule based systems and is independent of the application. The system permits the fusion of several sensors like optical, infrared, and SAR-images, laser-scans etc. and it can be used for the fusion of images taken at different instances of time. Sensor fusion can be achieved on a pixel level, which requires prior rectification of the images, on feature level, which means that the same object may show up differently in different sensors, and on object level, which means that different parts of an object can more accurately be recognized in different sensors. Results are shown for the extraction of roads from multisensor images. The approach for a multitemporal image analysis is illustrated for the recognition and extraction of an industrial fairground from an industrial area in an urban scene.