Advancing Image Analysis
Written by Richard W. Cooke
THE CHALLENGES OF PROVIDING ACCURATE
AND TIMELY GEOSPATIAL INTELLIGENCE
INFORMATION TO THE FIELD ARE BEING MET WITH
AUTOMATION, TECHNOLOGY AND INTEGRATION.
Military imagery and geospatial analysts face an ever-growing challenge in providing timely intelligence information to planners, analysts and field personnel. This challenge is driven by three key factors.
First, technology has made it easier for enemy combatants to detect surveillance and change locations, tactics and strategies much more quickly than in the past. This in turn is driving the requirement to decrease the time from data collection to field deployment of intelligence and situational awareness information.
The second factor is the rapid proliferation of imagery and GIS data sources, which has created a significant mismatch between the amount of data that must be analyzed and reviewed and the number of imagery and geospatial analysts available to review that data.
The third factor is that the ground resolution of these new datasets is so rich with information that it creates significant logistical issues in getting the geospatial information from the analyst to field-deployed personnel in a timely manner.
Further complicating each of these factors is the fact that imagery and geospatial analysis is still a complex science, which creates challenges in training new personnel to meet the growing demand for analysis of the increasing volumes of data. A natural tendency in the face of this complexity is to push for the simplification, consolidation and, to the extent possible, automation of the tools used to perform the analysis.
About six years ago, I sat across from an Air Force colonel who passionately promoted the need for a “find the tank” button in our software product, ENVI. As recently as May of this year, a colonel in the Indian Air Force said pretty much the same thing to me at a gathering of remote sensing and GIS professionals in New Delhi.
As much as the remote sensing software community would like to deliver such a one-button solution, the reality remains that the efficacy of the image science cannot be compromised just for the sake of simplicity. When you are working in a field in which accuracy counts and you depend on the science to deliver that, there is simply too much at stake. Users should beware of any software vendor claiming such a solution or even claiming that one software tool can do all things.
SIMPLIFICATION AND ACCURACY
There have been great strides, however, in the remote sensing software development community in the use of workflows and automation to solve some of these problems and reduce the complexity and overhead for analysts. Common tasks such as pan sharpening, land cover classification, feature extraction, change detection, anomaly detection and terrain categorization have become understood well enough across a broad enough sample of data types that vendors have successfully distilled these analysis tasks into automated and semiautomated workflows. This has significantly reduced the learning curve for analysts while also increasing the throughput of data that needs to be analyzed.
ENVI first introduced the automated workflow paradigm to users three years ago with its SPEAR (Spectral Processing, Exploitation and Analysis Resource) tools, which were very well received by analysts. Building on the success of SPEAR, we introduced an ENVI add-on module the following year, called ENVI FX, which provided a workflow-based approach to spatial and spectral feature extraction.Mp> Building on this concept, we will release a new product for image analysis this summer called ENVI EX. ENVI EX will combine the power of ENVI FX and SPEAR into a single package to provide GIS users with simplified, specific workflows for extracting information from imagery for use in GIS applications or to populate geodatabases for use in publishing maps and Web-based map services. ENVI EX will also offer analysts an unprecedented level of integration with the leading GIS platform, ArcGIS from ESRI.
The consolidation of software platforms is also a good approach to providing a consistent user interface to reduce workflow complexity, and thus improve productivity. Consolidation must be done with great care, however. Betting on one software package to meet all analysis needs is a dubious proposition at best, as it is not likely to meet the needs of analysts and forward-deployed personnel.
Recent initiatives in the community to develop a standards-based enterprise computing platform that allows best-in-class software from multiple vendors to work together, while providing a more consistent user workflow experience, is a more appropriate approach. This approach also does not preclude a predominantly COTS solution, the cost benefits of which are well documented. The extent to which vendors implement in accordance with industry standards improves the community’s abilities to deploy best-in-breed software within their enterprise that is best suited to various data types and mission needs.
The standards-based computing platform approach is one of true integration—using software development and industry standards to provide true and effective interoperability between software tools. This is the core motivation behind the strategic partnership we at ITT Visual Information Solutions have formed with ESRI to integrate our ENVI and IAS technologies with its best-in-class GIS platform, ArcGIS.
Beginning this summer, we will be releasing a series of ENVI products that will integrate seamlessly into the ArcGIS desktop and server software stacks in order to bring ENVI’s advanced image analysis tools and ESRI’s GIS tools into a single common workflow. Traditional ENVI users will benefit from having advanced GIS capabilities added to their workflow, while ArcGIS users will gain access to ENVI’s advanced analysis capabilities without having to switch between applications.
This will be further enhanced in mid- 2010, when this same integration will extend to the enterprise as ENVI functionality becomes accessible to users of ESRI’s ArcGIS Server and Image Extension.
NEW COMPUTING POWER
Matching these integration, algorithmic advancement and workflow developments with significant advances in computing power and performance will also help significantly in the improvement of productivity. Techniques such as machine virtualization to maximize the efficient use of computing resources and the use of graphics processing unit (GPU) processing to perform computationally intensive operations will greatly improve analysts’ ability to work through the volumes of data they are receiving.
Once again, vendors must develop to industry standards in order for users to take advantage of these advances in information technology standards and technologies. These techniques will grow increasingly necessary as datasets grow in resolution, and thus size, and as we move into the use of remotely sensed data from multiple modalities, such as synthetic aperture radar (SAR) and light detection and ranging (LiDAR), to perform fusion-based analysis. We are investing significantly at ITT in the development of both GPU processing techniques as well as in the fusion of EO, LiDAR and SAR data. Over the next year, users will see the results of this investment in our ENVI releases.
Once the analysis of the data is complete and the information products are created, the second half of the geospatial imagery challenge is to support a wide constituency that is geographically dispersed. Once again, the sheer size and amount of data involved is creating a logistical problem in terms of getting the data from analyst to the field.
The answer to this problem lies in compression techniques, but not necessarily in the manner one might think. Historically, compression was more about reducing the storage footprint of data rather than its transmission. But with the emergence of JPEG2000 as an efficient, lossless compression scheme, we can now stream imagery in real time, even over highly constrained communications pipes, while maintaining the lossless characteristics of the data.
This is accomplished by deploying ITT’s Enterprise Image Access Solutions (IAS), a COTS software solution that manages the compression as well as the encrypted, rapid streaming of geospatial data. Using a progressive download and rendering technique in IAS, users can drill down into areas of interest within an image, map or dataset in a matter of seconds rather than waiting hours for large files to transfer to their forwarddeployed image repository, if one exists.
Using asynchronous data fetching allows for progressively more data to be downloaded as the user pans and zooms through the imagery, and the new local context updates almost instantaneously.
ENTERPRISE SOLUTION
The Enterprise IAS solution, however, pre-supposes the existence of an imagery and GIS data store, or geodatabase in the case of more advanced users. This assumption is not always valid, particularly when dealing with forward-deployed teams that rely on centralized data repositories spread across multiple locations. While custom solutions to the storage and data management aspect of this problem abound, there is a clear opportunity to deploy COTS-based solutions for military and intelligence communities.
Working again with ESRI, ITT is integrating its Enterprise IAS software solution with ESRI’s ArcGIS Server and Image Extension to provide an end-to-end, off-the-shelf image cataloging, data management and dissemination solution. Based on service-oriented architecture (SOA) and Web services standards, this solution also provides the ability to quickly plug in analysis tools from ENVI to process data and imagery, and then publish data products to forward-deployed users as Web services—thus further enhancing the efficiency of data ingest to information dissemination.
These two technologies together will provide a solid foundation for users needing a range of imagery and geospatial analytical capabilities, while also helping them manage their ever-growing collection of data. The system can even be used to federate image catalogs from a variety of sources and locations to provide a seamless global view into multiple data stores, regardless of their location.
It is an exciting time to be in the software business, particularly as we tackle the very real and challenging problems of getting accurate and timely geospatial intelligence information to the field. To meet these challenges, ITT offers solutions that provide significant productivity gains across the workflow—from data ingest, storage and management to analysis and dissemination— in cost-effective COTS packages.
We are investing heavily in the distillation of complex analysis tasks into simplified workflows in our ENVI products to give analysts the tools needed to quickly and easily create increasingly in-demand imagery products. Working with our partners at ESRI, we are driving for integrated, standards-based platforms on the desktop and the enterprise so that analysts will achieve increasing gains in efficiency as our product lines converge into well-defined and seamless workflows. In addition, we are using the advances in computing technology, including processing, rapid streaming technologies and SOA architectures, to increase throughputs, management and dissemination of information products to field users. As always, we remain committed to supporting the critical geospatial intelligence mission. ♦
Richard W. Cooke is president of ITT Visual Information Solutions.