"There is one thing stronger than all the armies in the world, and that is an idea whose time has come." - Victor HugoThe availability of geospatial data sets is exploding. New satellites, aerial platforms, video feeds, global positioning system (GPS) tagged digital photos, and traditional geographic information system (GIS) information are dramatically increasing across the globe. These raw materials need to be dynamically processed, combined and correlated to generate value added information products to answer a wide range of questions.
This article provides an overview of OMAR (OSSIM Mapping Archive) web based geospatial processing. OMAR is part of the Open Source Software Image Map (OSSIM) project under the Open Source Geospatial (OSGeo) Foundation. The primary contributors of OSSIM make their livings by providing professional services to US Government agencies and programs. OMAR provides one example that open source software (OSS) solutions are increasingly being deployed in US government agencies. We will also summarize the capabilities of OMAR and its plans for near term development.
OMAR is a web based system for archival, retrieval, processing, and distribution of geospatial assets. Satellite and aerial images, vector sets, unmanned aerial vehicle (UAV) video sets, as well as user generated tags and reference items can be easily searched and manipulated with the system. Searching can be performed on the basis of location, time, or any combination of the stored metadata. OMAR is unique in its ability to dynamically process raw materials and create value added products on the fly. Imagery is orthorectified (geometrically corrected), precision terrain corrected, and histogram stretched on demand. OMAR combines, fuses, or chips areas of interest according to the users needs. Geospatial assets can then be manipulated, viewed, and processed to provide a wide range of value added products. These products are delivered through several mechanisms:
- a planned generated product that is distributed through ftp or email
- results generated with a simple browser interface
- open standards and interfaces such as the Open Geospatial Consortium's web map service (WMS), web feature service (WFS), and web coverage service (WCS), as well as tiling services (TileCache)
OMAR integrates several OSS solutions to provide an online dynamic processing solution. OpenLayers for dynamic mapping, PostGIS for the Postgres database, the GRAILS web framework, and OSSIM are a few of the technologies that are used.
The Open Source Business Model
Since 1999, OSSIM has evolved through US Government funding from the Defense and Intelligence communities. Throughout that period, the core development team has worked in a number of different companies while maintaining a close collaborative relationship. The combined OSSIM team has supported a number of different customers. As a result, OSSIM is now deployed in a number of critical government and commercial applications. Over time, a number of solutions and applications have evolved out of the core libraries. Solutions include:
- command line applications
- the ImageLinker prototyping tool
- ossimPlanet for 3D visualization and collaboration
- OMAR for online geospatial management and production
The advantages of an OSS approach are slowly becoming evident within the US Government. Changes in policy and acquisition practices are in work to spread the adoption of these practices. Successful projects such as OMAR and OSSIM are providing useful pathfinders as projects experiment and evaluate this approach.
The OSS business model is in the experimental stage within the US Federal government. The agencies are funding some professional services to extend and support OMAR, OSSIM and other OSS geospatial technologies. This model works well for organizations that manage prototype or development projects. Many agencies and organizations within the government do not directly support software development. Front line military operations, known as operational agencies, want certified, configured and supported solutions for their missions. Operational agencies tend to shy away from funding development efforts and a packaged product bundled with maintenance and support turns out to be a better fit for those types of agencies. However, it is becoming clear that open source solutions and support is being implemented across all levels of the government enterprise.
Given the level of interest in and exposure to OMAR, it is clear that a more traditional development team and support staff will be funded to meet operational and developmental requirements. Government mission requirements have identified a pressing need for additional functionality, an improved user interface, installation packaging, testing, documentation and support.
OMAR's relational database and geospatial processing capabilities allow rapid generation of value added information products from raw information. Delivery of this processed information can take the form of generated products, web browser views, or on demand web based services.
OMAR's cataloging and provisioning capabilities rapidly locate, process, and distribute value added products across the enterprise. OMAR can automatically detect, ingest, and process when new geospatial assets arrive in any of the monitored repositories. While there are a number of systems that can discover and distribute geospatial assets, OMAR is unique in its ability to process those assets into derivative products and services on demand.
The included OSSIM geospatial processing engine can construct image chains that define the functions, parameters, and conversions that are needed to read, re-project, and process the original geospatial assets into value added derivative products and services. Image chains are parameter driven instructions that describe how to build a value added product. These take the form of spec files that can be stored in the database.
Current installations of OMAR are managing millions of imagery and video files. Even though the current release is still considered in beta, the system is being used to find and rapidly view geospatial assets from multiple repositories. At this point, the primary user interface is through the browser as development begins to expose specific Web Services Description Language (WSDL) services and generated custom products.
A few of the current features of OMAR 2.0 are:
- discovery and online viewing of imagery and video
- national and commercial imagery as well as UAV video is actively ingested, stored and viewed in the system
- online browsing of imagery provided by OpenLayers and OSSIM
- playback of video clips currently provided by an external streaming service
- searching based on location, acquisition time, and metadata
- interfaces with the embedded PostGIS/Postgres relational database provide the ability to select assets based on location, time, or any of the values in the metadata including sensor types, target identifiers, and various collection criteria
OSSIM and Geospatial Data Abstraction Library (GDAL) provide support for a wide array of native geospatial formats. These libraries can provide file format conversion and allow assets to be referenced and used without intermediate conversion.
The OSSIM library provides on the fly reprojection and orthorectification for display and output. Satellite and aerial images are processed through OSSIM sensor models to map projected products.
Rigorous and Rational Polynomial Coefficient (RPC) sensor models apply precision terrain correction to Digital Terrain Elevation Data (DTED), Shuttle Radar Topography Mission (SRTM), or raster elevation data sets that are referenced to the system.
OSSIM creates complex products from spec files. The spec files define the parameters and processing steps needed to build a product. These image chains are being used in OMAR to define processing and views based on demand. Future development will include the ability to store user defined processes in the database for custom products.
Multi-image mosaics and fusions are being produced by the OMAR system in pre-defined image chains. New services and variations will be exposed in the future.
OSSIM and GDAL provide an extensive number of file format conversion services. The user interface needs to be extended to expose this for user generated products.
OSSIM dynamically re-projects data into geographic views. User selectable map projections and datums need to be exposed through the user interface. The underlying conversion process already exists. User defined areas of interest can be identified through the WMS interfaces. The underlying architecture has the ability to generate area of interest products and services for interfaces with external systems and user requests.
Much of the current development work is focused on exposing existing processing capabilities through the user interface, defining and implementing services, and developing targeted image chains for specific functionality.
Access through the system is authenticated through a user login, as seen in Figure 1. LDAP authentication is currently in development. Roles and privileges are granted based on the results. The current version of OMAR supports two levels of access: users and administrators. Once logged in, the user will be presented with an initial interface for browsing and searching.
Typically, the user will search for geospatial assets by selecting a geographic area of interest and filtering by acquisition date, sensor type, target identifiers, or any combination of the metadata tags that are stored in the internal database.
The administrator has the ability to add new tags and criteria to the search panel shown on the left in Figure 2.
The imagery search web interface, shown in Figure 3, provides a map that the user can pan or zoom to select a desired area of interest. Zooming the map to a particular area of interest reveals outlines of data sets that are available in the system. The user can then select the "Area of Interest" mode and draw a selection rectangle over the desired area for search.
Additionally, the fields in the left panel can be filled to further filter the search for data. A number of parameters are available for search criteria. The user can manually enter center and corner coordinates, acquisition time and date parameters, or values for any of the metadata tags. In many government applications this will include sensor ids, target identifiers, sensor types, or resolution criteria. Configurable metadata parameters and overviews of the data assets are displayed. Clicking on the overview thumbnail will allow interactive viewing of the full data set.
OMAR provides interactive zooming and panning into a satellite image. Behind the scenes, OSSIM is processing the raw file through a sensor model, cropping and zooming into the image, and enhancing the image with histogram stretching. Roaming, panning and zooming is accomplished interactively. The underlying image is being projected through a sensor model, orthorectified, precision terrain corrected and histogram stretched on the fly.
OMAR also can process UAV Predator feeds. These feeds can be searched, selected and played back through the web browser.
OMAR is able to extract and parse the metadata from Motion Imagery Standards Board (MISB) compliant video streams. Missions, acquisition dates and times, platform and center of interest coordinates are used to populate the internal database and position the data for geographic searches.
OMAR is an integration of several successful OSS projects to provide an enterprise solution for geospatial data management, production, and distribution. Defense and Intelligence agencies of the US government have provided funding to support OMAR and the underlying OSSIM software libraries. Through this support, the OSSIM development team has been employed through a number of collaborating projects.
Interest in OMAR is gaining across a number of agencies and it will soon evolve to a more formal government project. Maintaining OSSIM as an unclassified open source project on the Internet has been key to its success and its ability to collaborate across a number of separate government projects.
The OSSIM team is always looking for additional contributors, developers, and users. Additional information can be found at ossim.org.
OSSIM Whitepaper (pdf)