Remote Sensing is an important tool for the analysis and interpretation of a wide range of global and regional conditions and processes on the earth. Airborne and spaceborne Remote Sensing systems produce a rapidly growing number of data sets, and improvements in sensor technology result in continuously increasing spatial and spectral resolutions of these data sets.
To visualize Remote Sensing data for analysis and interpretation purposes, sensor data from various sources has to be processed and combined to produce geometry and color information. Since the details contained in multimodal high-resolution data sets cannot be preserved in a single static image, interactive visualization techniques are required.
A system for interactive visualization of Remote Sensing data allows to choose and adjust data processing and data fusion methods interactively. This enables the user to bring out the data details that are relevant for the current aims and objectives, and thus to gain better insight into the data.
Interactive visualization of Remote Sensing data is a challenging problem:
- Remote Sensing data sets are generally very large due to their high resolution and area coverage, and visualization tasks often need to combine multiple data sets. An interactive visualization system must be able to efficiently handle large amounts of diverse input data, and at the same time provide data processing capabilities that allow interactive adjustments.
- To produce geometry and color information from sensor data, specialized processing methods are required for different sensor systems. These specialized methods must allow interactive and intuitive adjustments.
- The dynamically generated geometry and color information resulting from interactive data processing and fusion must be rendered efficiently and accurately.
This dissertation proposes methods to address these challenges. A framework is presented that handles data management, processing, and fusion. Specialized interactive sensor data processing methods are examined based on the example of images generated by Synthetic Aperture Radar systems. Efficient geometry refinement methods are presented that are suitable for rendering dynamically generated data with guaranteed error bounds.