Automatic Extraction of Physiographic Features and Alluvial Fans in Nevada, Usa from Digital Elevation Models and Satellite Imagery through Multiresolution Segmentation and Object- Oriented Classification

There is a need to automate terrain feature mapping so that to make the process more objective and less time consuming by using proper feature extraction techniques. The objective of this study was the use of object-oriented image analysis methods for the automatic extraction of physiographic regions and alluvial fan landform units. The study area was located in Nevada, USA. The data used included an ASTER L1 satellite image, the 1 Digital Elevation Model and the GTOPO30 Digital Elevation Model, available by USGS. At first, a multiresolution segmentation algorithm was applied for extracting image primitives. A class hierarchy was defined in order to classify these primitives into semantic image objects. A fuzzy classification then provided the first approximations of three physiographic feature types (basins, piedmont slopes and mountains). Further processing, by a segment fusion technique, resulted in the reclassification of these image semantics into the final physiographic feature units. For the extraction of alluvial fan units, a multiresolution segmentation technique was developed, delivering object primitives at several resolution levels. At the finest level, the physiographic feature types were extracted from the DEM. At a medium level, a knowledge base including definitions of Alluvial Materials, Sediments, Basin Materials and Rock-Mountain Materials was implemented. This level was classified through several iterations, using spectral information for the first iteration of the classification procedure and heuristics concerning contextual information for the second iteration. Finally, at the coarse level, a projection was made, classifying the data into two classes: Alluvial Fans and Other Objects. The results were compared to manually produced maps by an expert geomorphologist and to computer-produced maps and they were found satisfactory.