You are here

Automated Feature Extraction Capabilities for the Development of High-Resolution GEOINT Feature Data and Constructing Correlated Databases

Award Information
Agency: Department of Defense
Branch: Special Operations Command
Contract: H92222-07-P-0006
Agency Tracking Number: S062-012-0114
Amount: $99,517.00
Phase: Phase I
Program: SBIR
Solicitation Topic Code: SOCOM06-012
Solicitation Number: 2006.2
Solicitation Year: 2006
Award Year: 2006
Award Start Date (Proposal Award Date): 2006-11-27
Award End Date (Contract End Date): 2007-05-27
Small Business Information
1525 Perimeter Parkway, Suite 325
Huntsville, AL 35806
United States
DUNS: 884102542
HUBZone Owned: No
Woman Owned: No
Socially and Economically Disadvantaged: No
Principal Investigator
 Sandra Vaquerizo
 Principal Investigator
 (407) 737-8800
Business Contact
 Mike Pavloff
Title: Director
Phone: (650) 224-5030
Research Institution

Our solution to the automated feature extraction problem will leverage the material properties that can be inferred from combining multispectral imagery with high resolution elevation data or LIDAR data using a trainable knowledge base. Multiple imaging bands provide a more complete picture of the material involved than ordinary RGB. This can help distinguish between a green grass lawn and a green concrete tennis court, providing more accurate feature identification. Adding elevation data will not only help find boundaries between objects, but can help bridge gaps in the imagery due to occlusion and help to characterize materials. For example, a tree may hide part of a road in the imagery, but elevation data procured using radar that penetrates trees or multiple returns from LIDAR could be used to verify the continuity of the road surface. The tool developed will be compatible with the Common Database (CDB) structure of tiles that consist of multiple layers, containing different geospatial information.

* Information listed above is at the time of submission. *

US Flag An Official Website of the United States Government