|
Track: Defense and Intelligence
Steven Cento
Northrup Grumman Corporation/ATDC
South Oyster Bay Road
M/S K07-14
Bethpage, NY 11714-3584
Telephone: 516-346-8831
Fax: 516-575-0965
E-mail: stevec@tech.grumman.com
James McNeely
Automated Conflation and Update of Geospatial Feature Data Bases
Defining Issue: Automatically maintain a unified digital geospatial natural/cultural feature database that can provide the war fighter with a common, consistent, and current view of the battlespace in support of mission requirements. GIS Solution: Use an integrated hybrid of advanced technologies (i.e., artificial neural networks, fuzzy logic, evidential reasoning, etc.) to help automatically integrate, reconcile, and update multiple disparate digital geospatial feature data sources/databases in near real time. Implement a unified vector Data Base (UVDB) using an open geographical interchange standard (OGIS)-compliant software isolation wrapper around lower-level calls to the Esri Spatial Database Engine (SDE) to support attributed geospatial data storage, retrieval, query, and update functional primitives. Once a reconciled baseline UVDB has been constructed from separate National Imagery and Mapping Agency (NIMA) and other vector data sources, the timeliness of this database is maintained using more
dynamic intelligence, aeronautical, and imagery update sources. Methodology: The Air Force Rome Laboratory Multiple Data Base Integration and Update (MDBI&U) system automatically maintains the currentness and consistency of a digital geospatial feature database using a variety of update source materials that are commonly available within Air Force channels: (1) intelligence message traffic and intelligence databases (i.e., United States Message Text Format [USMTF], the Modernized Intelligence Data Base [MIDB], and so forth); (2) monthly aeronautical update sources (i.e., NIMA Electronic Chart Updating Manual [ECHUM] hazards to air navigation, Digital Aeronautical Flight Information File [DAFIF], and so forth); (3) primary image sources (e.g., reconnaissance photos, satellite imagery) and secondary image sources (e.g., Image Product Archive [IPA] push/pull server); (4) digital terrain (e.g., NIMA Digital Terrain Elevation Data [DTED], custom-produced elevation information) and other update source material
categories. The baseline UVDB is produced by automatically performing conflict resolution (i.e., conflating) on NIMA and other digital vectorized natural/cultural feature coverage sources of disparate format, content, attribution, scale, accuracy, confidence, and/or distribution media. Advanced technologies like Artificial Neural Networks (ANN) are used to detect and resolve the myriad of pattern assimilation, completion, and matching problems that typically arise during conflation and update of geospatial vector data. These same technologies are equally appropriate for updating the baseline UVDB. A three-step "update loop" (composed of Candidate Retrieval, Candidate Selection, and Database Update cooperating subprocesses) is carefully designed to interchangeably support both vector data set coverage conflation (of a batch-type nature) and "on-the-fly" database updates (of a piece-meal nature) using "the same" software. A unified all-encompassing database schema is established based on (1) an extended
Digital International Geographic Exchange Standard (DIGEST) Feature Attribute Coding Catalogue (FACC) that augments NIMA feature attribution to include intelligence, aeronautical, private sector, and other vector input/update sources; (2) a Vector Product Format (VPF) compatible database schema supporting creation, storage, and maintenance of nodal and directional topology; and (3) export of left/right polygon face topology built "on-the-fly" from UVDB contents. An attribute thesaurus is used to resolve conflicts between disparate attribution standards (e.g., NIMA versus Defense Intelligence Agency [DIA] attributes). Software: The core MDBI&U system was developed using ANSI C, X-Windows Version 11 Revision 5, and Integrated Computer Solutions (ICS) Motif 1.2.3 on a Solaris 2.x UNIX operating system (OS) and Sun SPARCstation 10/20 workstation platform. An integrated hybrid of government-off-the-shelf (GOTS) message handling systems is used to support USMTF intelligence message parsing and reformatting in
a neutral update format (NUF) to generically support update content extraction. The MDBI&U database was developed using Esri SDE/Oracle, a streamlined, applications programmer interface (API)-callable back-end commercial-off-the-shelf (COTS) spatial database engine that integrates high-powered spatial and relational capabilities within a singular client-server based data handling environment. SYBASE is also used to support intelligence database interfaces (e.g., MIDB). The purpose of the paper is to highlight the MDBI&U system and its functions, as well as to review the various technical challenges encountered during automated vector conflation and all-source updating of a unified vector data baseline. The problems detailed include technical algorithm and SDE applications development and systems integration issues.
|
|
