I partnered with Ducks Unlimited, Inc. to update the National Wetland Inventory (NWI) database for a 7,200 square mile area surrounding the Twin Cities.  I provided image segmentation and data integration services that were used by the wetlands specialists at Ducks Unlimited to speed up the NWI mapping process. This combination of experienced photo-interpreters and cutting edge image processing technology reduced the labor required to complete the project by approximately 25%. The primary imagery for this project was 50cm resolution color/infrared aerial orthophotography.  I developed an automated image segmentation process to identify key wetland boundaries in the imagery. By using the segments created in eCognition, the image interpreters could spend much less time drawing lines. This allowed them to focus their attention on the more difficult task of classifying wetlands by type.  Many sources of information, including LiDAR derived DEMs, SSURGO soils information, SONAR bathymetry data, and PALSAR satellite RADAR, were available to assist the interpreters in determining wetland type. I helped to organize this information by integrating information from other geo-spatial data directly into the image segmentation process.  The end result was a high quality, up-to-date National Wetland Inventory database.  |
Project Portfolio >