Associate Professor (Research) in Spatial Sciences and Informatics
Yao-Yi Chiang received his Ph.D. degree in Computer Science from the University of Southern California in 2010; his Bachelor degree in Information Management from the National Taiwan University in 2000. His general area of research is artificial intelligence and data science, with a focus on information integration and spatial data analytics. He develops computer algorithms and applications that discover, collect, fuse, and analyze data from heterogeneous sources to solve real world problems. Dr. Chiang is also an expert on digital map processing and geospatial information system (GIS). He has published a number of articles on automatic techniques for geospatial data extraction and integration.
Prior to USC, Dr. Chiang worked as a research scientist for Geosemble Technologies and Fetch Technologies in California. Geosemble Technologies was founded based on a patent on geospatial data fusion techniques, and he was a co-inventor. Geosemble Technologies was acquired by TerraGo and Fetch Technologies was acquired by Connotate, both in 2012.
2005 - Present
Let's meet at conferences
I am a regular attendee of the ACM SIGSPATIAL International Conference on Advances in Geographic Information Systems (ACM GIS), International Conference on Geographic Information Science (GIScience), International Conference on Document Analysis and Recognition (ICDAR), IAPR International Workshop on Graphics RECognition (GREC), and International Conference on Pattern Recognition (ICPR).
Publication frequency by venues
HARVESTING GEOGRAPHIC FEATURES FROM HETEROGENEOUS RASTER MAPS
Yao-Yi Chiang, Ph.D. Dissertation, University of Southern California, Sept. 2010 PDF
Raster maps offer a great deal of geospatial information and are easily accessible compared to other geospatial data. However, harvesting geographic features locked in heterogeneous raster maps to obtain the geospatial information is challenging. This is because of the varying image quality of raster maps (e.g., scanned maps with poor image quality and computer-generated maps with good image quality), the overlapping geographic features in maps, and the typical lack of metadata (e.g., map geocoordinates, map source, and original vector data).
Previous work on map processing is typically limited to a specific type of map and often relies on intensive manual work. In contrast, this thesis investigates a general approach that does not rely on any prior knowledge and requires minimal user effort to process heterogeneous raster maps. This approach includes automatic and supervised techniques to process raster maps for separating individual layers of geographic features from the maps and recognizing geographic features in the separated layers (i.e., detecting road intersections, generating and vectorizing road geometry, and recognizing text labels).
The automatic technique eliminates user intervention by exploiting common map properties of how road lines and text labels are drawn in raster maps. For example, the road lines are elongated linear objects and the characters are small connected-objects. The supervised technique utilizes labels of road and text areas to handle complex raster maps, or maps with poor image quality, and can process a variety of raster maps with minimal user input.
The results show that the general approach can handle raster maps with varying map complexity, color usage, and image quality. By matching extracted road intersections to another geospatial dataset, we can identify the geocoordinates of a raster map and further align the raster map, separated feature layers from the map, and recognized features from the layers with the geospatial dataset. The road vectorization and text recognition results outperform state-of-art commercial products, and with considerably less user input. The approach in this thesis allows us to make use of the geospatial information of heterogeneous maps locked in raster format.
Ching-Chien Chen, Craig A. Knoblock, Cyrus Shahabi, Yao-Yi Chiang, Automatically and Accurately Conflating Road Vector Data, Street Maps, and Orthoimagery United States Patent 20070014488
I teach subjects in Spatial Sciences, Computer Science, and Data Informatics at University of Southern California
SSCI-582, USC Geographic Information Science and Technology (GIST)
SSCI-586, USC Geographic Information Science and Technology (GIST)
CSCI-599, USC Computer Science
SSCI-592, USC Geographic Information Science and Technology (GIST)
INF-553, USC Informatics
I collaborate with Profs. Craig Knoblock and Pedro Szekely at USC Information Sciences Institute on a number of research projects. Here's some information about our current projects.
For more details, take a look at our group page at Spatial Sciences Institute here.
Spatial technology opens a window into history
Best Vision Paper, First Place (2015 ACM SIGSPATIAL)
Certified GIS Professional, 2015
Best Paper Award, Second Place, the Forth Annual Intelligent Systems Division Graduate Student Symposium, 2008
Best Paper Award, Second Place, the Forth Annual Intelligent Systems Division Graduate Student Symposium, 2009
The Viterbi School Doctoral Fellowship, 2007 - 2010
I have been working with USC undergraduate and graduate students (in Computer Science, Geographic Information Science and Technology, and GeoDesign) on credit or no-credit direct research effort. Please feel free to contact me if you are interested to work on one of our projects.
Take a look at our group page here.
We'll be in touch real soon!