Harvesting Geographic Features from Heterogeneous Raster maps

The Faculty of Informatics is pleased to announce a seminar given by Yao-Yi Chiang

TITLE: Harvesting Geographic Features from Heterogeneous Raster maps
SPEAKER: Yao-Yi Chiang, Ph.D. student, University of Southern California
DATE: Thursday, July 30th, 2009
PLACE: USI Università della Svizzera italiana, room SI-006, Informatics building (Via G. Buffi 13)
TIME: 15.30

ABSTRACT:
Maps are widely available for areas around the globe and are an important source of geospatial information. Due to the popularity of Geographic Information System (GIS), high quality scanners, and Internet, we can now obtain various maps in the raster format. Comparing to other geospatial data, raster maps are easily accessible and provide geographic features that are difficult to find, such as currently inexistent landmarks in historical maps. Moreover, for certain types of geographic features, raster maps contain the most complete set of data, such as the USGS topographic maps that have the contour lines of the entire United States in various scales. We can use the raster maps to provide additional knowledge for viewing and understanding the other geospatial data. Furthermore, we can produce map context by extracting layers of geographic features (e.g., images of the text layers in raster maps) and recognizing the features (e.g., text labels) from raster maps. The map context then can be used for indexing and retrieval of the maps and the geospatial data that are aligned to the raster maps. For example, we can align raster maps to imagery and annotate the imagery, such as providing the terrain information by aligning a contour map to an imagery.

Harvesting the geographic features in raster maps is a challenging task for a number of reasons: First, maps are complex and normally contain overlapping layers of geographic features, such as roads, contour lines, labels, etc. Second, the image quality of raster maps is sometimes poor due to the scanning and/or image compression processes for preparing the maps in the raster format. Third, the access to auxiliary information of raster maps is often not available, such as the vector data used to produce the raster maps, map geocoordinates, legend information, etc. To overcome these difficulties, this talk presents both automatic and supervised approaches that utilize image processing and graphic recognition methods to exploiting the geographic features in heterogeneous raster maps of various complexities and image qualities. For raster maps with good image quality (e.g., digitally generated raster maps from vector data), I present an automatic technique that requires no user input for decomposing raster maps into images of geographic features (i.e., feature layers). For raster maps with poor image quality, I present a supervised technique that includes user training to extract the feature layers from raster maps. The supervised technique minimizes user input and utilizes image classification methods to reuse the user training results for reducing human effort. Lastly, I present feature recognition techniques that recognize the geographical features from the extracted feature layers. The overall technique described in this talk enables us to exploit the geospatial information in divergence maps.

BIO:
Yao-Yi Chiang is currently a Ph.D. student at the University of Southern California. He received his M.S. degree in Computer Science from the University of Southern California in December 2004. His research interests are on the automatic fusion of geographical data, image processing, and machine learning.  He has wrote and co-authored several papers on automatically fusing map and imagery as well as automatic map processing.  Prior to his doctoral study at USC, Yao-Yi worked as a Research Scientist for the Information Sciences Institute and Geosemble Technologies.

HOST: Prof. Fabio Crestani