Executive Summary Imagery Intelligence (IMINT) will be a mainstay of the Intelligence Community (IC) in the 21st century. The IMINT community (IMC) today is made up of a diverse set of users including military, national, and civilian. We anticipate that the numbers and types of imagery users will continue to grow dramatically in the future, perhaps into other areas not yet imagined. Thus, it is extremely important that our imagery system be flexible to support these changing needs. Exploitation will be the chokepoint for the imagery community. Given present trends, the number of images collected will continue to outpace our ability to analyze them. Collection costs continue to rise at the expense of processing and exploitation. Imagery analysts are working with archaic tools and the current acquisition process does not facilitate the timely infusion of new technology. This is due in part to the fragmentation of the imagery community, with infrastructure and research and development being pursued by numerous organizations with little to no coordination. Commercial imagery needs to be considered as an adjunct to national systems and plans must be put in place to facilitate its use. The IC continues to move to a dichotomy in imagery requirements: users want images in near-real-time yet also want detailed analysis. The imagery community has not yet reconciled how to satisfy these conflicting requirements concurrently. Imagery dissemination to the military below the Joint Task Force level still remains an issue and, finally, foreign denial and deception activities continue to be a problem and must be taken into account in future planning. IMINT will see a great transformation in the next century. Commercial systems will allow everyone, including our foes, to have access to high resolution imagery. At the same time, classification of national imagery must provide the required access to allies while continuing to protect collection/processing capabilities. More cost effective collection systems are required to free up funding to support the "downstream" activities of processing and exploitation. The explosion of available imagery requires that new technologies and exploitation/production tools such as automated/assisted target recognition algorithms and digital softcopy search tools must be aggressively developed to help streamline the exploitation process. The IC must move to all-digital exploitation of imagery, with access to cross-INT databases, while progressing to a "virtual" analytic environment, and funding must be increased to accelerate the procurement of softcopy (digital) workstations for imagery analysts. Support for the National Technology Alliance should be increased to provide more flexibility in rapidly fielding new technologies and exploiting commercially available technologies. Finally, increased emphasis should be placed on spectroradiometric collection, processing and exploitation. Thus, there is much in store for the IMC; however, it will not come for free. Funding must be increased to set up the central infrastructure needed to support the diversity of analysts, to bring those analysts the tools they need to help alleviate the exploitation chokepoint, and to increase and focus the R&D efforts to bring new technology to bear in a more rapid manner. Collection costs must be reduced so next generation systems and exploitation advances can occur. If these things do not occur, the IMC will not be able to satisfy 21st century requirements. IMINT: Imagery Intelligence Overview Imagery Intelligence (IMINT) will be a mainstay of the Intelligence Community (IC) in the 21st century. The IMINT community (IMC) today is made up of a diverse set of users including military, national, and civilian. We anticipate that the numbers and types of imagery users will continue to grow dramatically in the future, perhaps into other areas not yet imagined. Thus, it is extremely important that our imagery system be flexible to support these changing needs. The needs of the military will continue to expand, as their mission spreads into new, uncharted areas. Across all levels (strategic, theater, and tactical) we will see this new scope, in areas such as coalition operations; highly mobile, detached operations; enhanced C4I (Command, Control, Communications, Computers and Intelligence); and peacekeeping and humanitarian operations, along with further "operations other than war." These increased areas of responsibility bring with them a greater need for imagery support. Advanced, precision guided munitions will also demand a new level of sustained, highly accurate, imagery products. Civilian and national imagery requirements will also continue to grow. We have already seen the use of national imagery spread into environmental monitoring and evaluation and aid in disaster relief, both national and international. Nevertheless, this particular intelligence source will be of primary importance for support to law enforcement, counternarcotics and counterterrorism, monitoring treaties and weapons proliferation, and strategic and economic intelligence. Again, though, there may be areas of intense, future civilian use that go unseen today, because the future availability of commercial imagery and the recent push to downgrade national imagery will potentially bring out new and different users who did not previously have access to this type of data. Consequently, our future systems must be easily adaptable in order to meet these vastly different requirements. FINDING: IMINT will continue to be an important collection discipline for a wide variety of issues: indications and warning; support to the military; and monitoring arms control agreements, refugee flows, narcotics cultivation and ecological problems. The IMC faces several challenges and must adapt in order to maintain the level of support provided to, and expected by, today's customers in a future, changing environment. These challenges arise in almost every functional area: organization, requirements management, collection, tasking, processing, exploitation, and dissemination. Other issues include classification levels, denial and deception, and interaction with commercial systems. Each of these areas will be addressed separately in this study. IMINT will see a great transformation in the next century. Commercial systems will allow everyone, including our foes, to have access to high resolution imagery. At the same time, classification of national imagery must provide the required access to allies while continuing to protect collection/processing capabilities. The number of users and requirements will grow. Exploitation will be the chokepoint in the imagery process. The explosion of available imagery will overwhelm the imagery analyst unless automated/assisted target recognition algorithms or other exploitation/production tools can be developed. Spectroradiometric collection will become more important, with major impacts on the collection, processing, dissemination and exploitation arenas. Thus, there is much in store for the IMC; however, it will not come for free. Funding must be increased to set up the central infrastructure needed to support the diversity of analysts, to bring those analysts the tools they need to help alleviate the exploitation chokepoint, and to increase and focus the R&D efforts to bring new technology to bear in a more rapid manner. Collection costs must be reduced so next generation systems and exploitation advances can occur. If these things do not occur, the IMC will not be able to satisfy 21st century requirements. Organization Much attention has been paid to the IMC's organization in recent months. However, great care must be taken not to break those parts that work well in an attempt to fix other perceived problems. It is obvious that the current Central Imagery Office (CIO) does not have the authority it needs to oversee a diverse imagery community. Yet, before we rush into a new organizational structure, we must ensure that this new organization, while solving immediate problems, will be flexible enough to cope with the next century's "virtual" intelligence environment. FINDING: CIO does not have the required authority to oversee and effectively manage the imagery community. We are most concerned about a lack of CIO's authority to oversee an imagery strategic plan. Current imagery organizations are not tied together nor beholden to such a strategic plan. This results in disparate, uncoordinated allocation of funds and resources in collection, R&D, and exploitation and dissemination infrastructure. Dissemination within theater is another area that needs drastic improvement. Those areas that work well, though, are mainly within the exploitation community. Exploitation support to the policymakers is excellent. Support to the military is also very good in the areas of strategic indications and warning, and contingency planning. However, providing adequate imagery support to on-going operations is still a challenge, and will only be more difficult in the future. Thus, it is important for any new organization to look at this picture and show how deficiencies will be improved while maintaining the strengths of the previous organizations; at the same time, this new organization must be considered within the wider context of the IC. FINDING: The imagery community is badly fragmented. Infrastructure and R&D are being pursued by numerous organizations with little or no coordination. However, any restructuring should be considered only within the wider context of all other intelligence functions and activities. Some have suggested that a new organization be fashioned after the Signals Intelligence (SIGINT) model. Though it appears to be a convenient organizational structure, we do not believe this will solve the IMC's problems because the analogy of the National Security Agency (NSA) is not directly applicable to imagery due to major technological and operational differences in the two disciplines. We are also concerned that a major monolithic agency will be LESS responsive rather than more responsive to the customer. Finally, the risk that future imagery systems will be driven solely by technology rather than users' needs increases under these proposals (though this danger does exist with today's organizations). Some also claim that another major raison d'etre for this new organization is to solve the dissemination problems of DESERT STORM. We overwhelmingly agree that dissemination is a problem; however, it is hard to comprehend how an organization that has no control over theater/Joint Task Force (JTF)/Joint Intelligence Center (JIC) level forces and lower echelons will be able to solve this problem. Thus, we must again gravitate to the real problems within the IMC and focus on an organization that will be able to provide solutions to these problems. The main problem areas we see with the current structure are imagery program management/planning, research and development (R&D), collection, processing, dissemination, and standards. A single, strong policy arm is needed for coherent end-to-end planning. Several key functions should be centralized: standards, protocols, and communications interfaces. A strong R&D oversight structure must be included to ensure that new technologies are responsive to customer requirements and that R&D funds are spent efficiently, according to an overall plan instead of each organization funding bits and pieces as is done today. The IC21 Intelligence Community Management staff study presents an IC that will solve these deficiencies through the needed centralization of certain functions while preserving those areas that work well. We believe the exploitation community is one of those areas. This is an area where IMINT differs greatly from SIGINT. In the SIGINT arena, a signal is collected and analyzed by NSA, producing information which is then distributed to a variety of customers and agencies. IMINT, on the other hand, produces an image which is then sent to a variety of organizations and exploited in many diverse ways within those organizations. Hence, imagery exploiters are, in many ways, discrete customers/users of the imagery in and of themselves and, thus, the SIGINT analogy is really not applicable in this case. Keeping imagery analysts close to their customers will become increasingly important but too great a dispersion of capabilities may lead to an erosion of imagery analysis expertise. Thus, a balance must be struck between decentralization and centralization of imagery analysis capability. Another balance that must be struck is the level of segregation between military analysts, analysts who support national and civilian customers, and cartographers. Recent recommendations have been to combine these forces into one exploitation cadre. Again, we go back to our argument that the different exploitation elements should be treated as discrete customers. There is danger in too much centralization because of the diverse sets of skills these analysts bring to the table. We fear that combining these personnel into one homogenous unit will dilute these skills into one set of "accepted" skills, which will not completely satisfy any customer's requirements. In order to preserve the diverse set of analytical skills we have today, we recommend keeping the disparate imagery analysts with their originating parent organizations, while centralizing the infrastructure that supports them; however, we also recommend better integration of the imagery analysts into those organizations for better support to the "all-source" analysts. RECOMMENDATION: As noted in the Intelligence Community Management staff study, second and third-tier analysts from all INTs should be co-located with true "all-source" analysts in the CIA and DIA. We must look to the future, not the past, for a new organizational model. Legacy stovepipe organizations are a product of the past and will not provide the needed flexibility required to support a "virtual" intelligence community in the future. Our model of IMINT in the 21st century is based on centralization of vital functions (end-to-end planning/management, R&D, collection, processing, archiving, and infrastructure) while sustaining a diverse customer/exploitation base. Needs of the users must drive the organization and those users' needs are met mainly by imagery derived information and products prepared by professional imagery analysts, not the raw image. These decentralized production strengths equate to increased responsiveness to local needs/missions and the ability to tailor and/or focus efforts quickly to respond to changing priorities. This flexibility in exploitation, combined with consolidation of programmatic and tasking oversight, and a standards based infrastructure, will truly allow the IMC to be responsive in the 21st century. Requirements Requirements on a grander scale and collection synergy are discussed in separate IC21 studies. However, in the context of imagery, requirements and collection management must be discussed. The new Requirements Management System (RMS) for imagery is due to reach initial operating capability (IOC) in June 1996 (eighteen months behind schedule). It is unclear at this time whether RMS will be able to perform comparably to its predecessor, CAMS (COMIREX Automated Management System). In all fairness, the RMS goal was admirable: to allow the user to follow his imagery request and know exactly where it was in the requirements process. However, it is a possibility that RMS will never be able to achieve full operating capability. This is a great example of spending large sums of money on a stovepipe system. Of course, it was expected that this system would be up and running by now, and that we would be on our way to designing the collection management tool of the future. Since this is not to be, in the near-term, we must ensure that RMS will provide equivalent capability before we allow CAMS to be shut down. (Both systems cannot run simultaneously.) In the event RMS cannot meet expected performance levels, CAMS must be retained until the next generation system is available. For that next generation system, we envision an integrated requirements process where all types of intelligence collection are tasked (e.g., SIGINT, IMINT, MASINT, etc.) Ideally, this translates into one requirements tasking system. The military's Joint Collection Management Tool, which was supposed to interface with RMS, is a small step in the right direction and provides only one interface to the process. However, this is not absolutely necessary. What is required, though, is consolidated resource planning. We must be able to do cross-platform, cross-sensor tasking, with dynamic and flexible planning, scheduling, and management. Managing which users get to steer which collection assets will be difficult. Rapid exploitation feedback will allow more optimized planning and scheduling. This all-source requirements system must be compatible with theater/tactical assets and should look to meet the goals set out by RMS, mainly that the customer would know the status of his request, for all -INTS, throughout the entire process. This is discussed in much further detail in the other staff studies mentioned above. Validation of imagery requirements also needs an overhaul. The current Community Imagery Needs Forecast (CINF) does not currently include all requirements. It also appears that requirements are based upon what collection systems are/will be available instead of what information is required. It appears that the "Seal of Approval" process does not address cost effectiveness or ability to fulfill requirements. FINDING: the CINF is incomplete and appears to reflect only what can be collected versus what needs to be collected. We need a requirements system that is immune to special interests. We propose a central requirements organization that would look across all -INTs to determine the most cost-effective and capable way to collect the required information. We need an organization that looks to the future to determine which technologies require increased investments today. We would like to see the IC study and react instead of study and report. There must be thorough understanding of the problem before the IC jumps to solutions. But this should take months, not years. RECOMMENDATION: As noted in the Intelligence Community Management staff study, a Community Management Staff with IC-wide authority over requirements, resources and collection would improve the role of all collection disciplines. This would also abet a more integrated requirements and tasking system for IMINT, which has yet to be attained. Collection FINDING: Collection costs continue to rise at the expense of processing and exploitation. Only one solution has been offered so far that shows major promise in reducing costs while maintaining capabilities: small satellites (smallsats) acquired through streamlined acquisition practices. A distributed architecture made of smaller, single function satellites, will provide the flexibility and responsiveness required for the customers of the next century. Technology is now available that would allow the IC to shrink its satellite size, thus reducing costs, both for the satellite and the launch vehicle, but also from an organization infrastructure point of view. Also, by using streamlined acquisition, this approach allows new technology to get on-orbit more quickly. Multispectral sensing satellites can be added to supplement this architecture. Best commercial practices must be incorporated. RECOMMENDATION: Move to an architecture of small satellites (smallsats) to increase capability, flexibility and revisit while reducing costs. Smallsats have also been proposed for point targets that need high resolution collection. These Narrow Field of View (NFOV) satellites, while more complex than the Wide Field of View (WFOV), offer an exciting opportunity to maintain capability but at much reduced cost. Unfortunately, because many people believe smallsats are only capable of fulfilling narrow, niche missions, these types of satellites will never be considered seriously until this technology is proven on-orbit. (It appears that it is more widely accepted that the WFOV mission can be done than the NFOV mission). Therefore, we must build and fly a NFOV small imager to convince the skeptics that we do not need to spend billions per satellite to have equivalent capability. Thus, we must act now. As stated earlier, smallsats will not be considered as a viable alternative unless there is an on-orbit demonstration showing their worth. It is imperative that the small NFOV satellite be built as quickly as possible in order for this technology to be a serious contender. RECOMMENDATION: Proceed quickly with a small satellite demonstration in order to ensure this option is considered as a viable alternative for the next generation of imagery satellites. Another idea that should be reviewed, especially if the cost per satellite can be contained, is to reverse the trend of increasing Mean Mission Duration (MMD) and build satellites that will last only three to four years. Costs would be further reduced, both per launch vehicle and satellite, because larger block buys of both systems would allow a cheaper unit price. Limiting the lifetime of satellites would also allow advanced technology to be incorporated more quickly and missions to be altered to adapt to new situations because satellites would be replaced at a relatively fast pace. Industrial base concerns would be alleviated and launch crews would always be current on their procedures. The recent push to increase MMD seems to be a survival tactic to counter the large growth in satellite cost; because the IC's satellites have grown so expensive, we can buy only a few, spaced out over several years. Thus, these satellites must last longer so the IC can stretch out its costly acquisitions. This approach should be given closer scrutiny. FINDING: "Denial and deception" activities by foreign governments are a current problem. As U.S. imagery capabilities become more widely known, this problem will likely grow. Along these lines, the exploiters should be viewed as customers and as such should have input in deliberating the value of new systems because they are the ones who must use the product. They should have direct involvement in utility studies of new types of systems, which is not the currently the case. Today, the National Exploitation Lab (NEL) is only involved in these types of studies when asked to participate. They, along with all other primary users, should have the authority to demand involvement with the evaluation of any new imagery system. The same "clean sheet" argument can be made for the command and control and ground processing segments for imagery. New commercial satellite architectures will be required to control on the order of hundreds of satellites. Can we leverage off of the work they are doing? New processing advancements are being made in the commercial sector that should be incorporated quickly. This appears to be only one of many examples where contractors have conveniently made themselves indispensable, at the expense of the government. FINDING: The current imagery ground architecture is very complicated and expensive. Commercial companies are developing ground stations at much less cost. It gets back to the principle of deleting unnecessary layers and overlapping influences, wiping the slate clean and starting over. The IMC should look at using the "clean sheet" approach for its ground functions. It is especially important that this method be implemented now while the "lessons learned" expertise still resides within the NRO. Thus, they would have the advantage of quickly infusing new technology and simplifying operations while ensuring that mistakes of the past are not repeated. RECOMMENDATION: Redesign the ground architecture from a clean sheet of paper in order to take better advantage of commercial capabilities and reduce ground station vulnerabilities. The NRO needs to return to streamlined program offices with smart people doing the work, thus reducing the need to rely on numerous SETA and support contractors. This, too, will reduce the costs required to procure satellites. On the airborne side, Unmanned Aerial Vehicles (UAV) and airborne collection will continue to be important assets to support the theater and tactical commander. However, their collection capability is limited to only those areas where they can fly with impunity. However, for all airborne collection that remains, the imagery must be collected digitally in order to ensure its compatibility with future imagery databases and exploitation workstations. The tasking of these systems should be integrated with the tasking of overhead systems in order to maximize efficiency and delete duplication of collection. Exploitation/Information Processing Exploitation will be the chokepoint in the imagery process of the future. The amount of imagery collected will be increased greatly at a time when the number of imagery analysts will have been reduced. How to interpret new types of imagery like multispectral collection will have to be learned at a time when it will be impossible to pull analysts off-line, unless the hiring trend for analysts is reversed. Softcopy workstations are a critical need and purchases for all imagery analysts should be accelerated. These workstations should be compatible or able to be upgraded to work with all types of intelligence and their associated databases. R&D for a softcopy search tool should be a number one priority. Either the number of analysts must be greatly increased or technology must be developed to make both the analyst workforce more efficient and to take away some of the exploitation preparation workload. We would venture that both must occur: the number of analysts must be increased and the technology must be developed, both in the forms of better workstations and better tools. R&D dollars must be consolidated in order to better serve the imagery community; however, each organization must have control over some amount of funding in order to preserve specialized tools. FINDING: Given present trends, the number of images collected will continue to outpace our ability to analyze them. FINDING: Imagery analysts are working with archaic tools; the current acquisition process does not facilitate the timely infusion of new technology. The number of analysts needs to increase now. Also, we are facing a severe deficit down the road because of a reduction in the number of imagery analysts. The longer we wait to begin rehiring, the greater the danger we will face a gap in knowledgeable imagery exploitation. Fifty percent of DIA's imagery workforce will be eligible to retire within the next five years. This is a problem that cannot be ignored because it takes several years to train an imagery analyst to be self-sufficient. Another problem that has occurred because of downsizing is the "in-box" mentality. This is not just a problem within the IMC but is occurring everywhere within the IC. Analysts are too busy dealing with the crises of today to have the time to think creatively and look long range. DIA, in the past, apportioned part of their personnel to look at long-term issues but they no longer have this capability. History shows that there will be problems that may take interdisciplinary teams years to solve. With the current emphasis on immediate information, there is a danger that refined, thoroughly analyzed intelligence will become a thing of the past. We must balance real-time information needs while protecting long-term research. Another issue is the availability of analysts for the testing of new tools, products, etc. It is currently very difficult to pull analysts off-line for this purpose because there is no margin left in the number of analysts doing the day-to-day work. All of these problems hinge on the number of available analysts. Hence, we must act quickly to increase the number of imagery analysts, both national and military. The optimum number of analysts will depend greatly on the exact mission the IMC is asked to perform and on how well we apply technology to streamlining the exploitation process for those analysts. Regardless, the number we have today is inadequate and, due to the long timelines of training, hiring of new imagery analysts should commence at once. Future imagery analysts will face even harder tasks. They will be required to look at and evaluate diverse types of imagery and use more sophisticated tools. They will also work daily with a paradox: producing thoroughly analyzed, contextually based products while meeting demanding timeliness requirements of less than 24 hours (in some cases, 12 hours). This is an impossible task in today's environment, yet will become increasingly more important in the future as other countries gain access to similar imagery. Strategic advantage will become a matter of whose collection, exploitation and dissemination timelines are the shortest. Intelligence must be there swiftly so as to be relevant to decreasing planning and execution timelines, and packaged in such a way that can be consumed by the user. The lower echelons of the military present the real crux of the problem: extremely short timelines must be met yet great detail is still required. This would appear to be a push toward automated exploitation; this however, implies that the time-dominant reporting will not have analyst derived information and will merely report what, where and when, not who, from where and why. In some instances, this may be all that is required but it is our belief that a human will always be needed, at least during the timeframe dictated for this study, to provide the cognitive processes of exploitation. Nevertheless, R&D should be increased and focused on providing these analysts the new tools and efficient processing capability required to help them come closer to meeting these demanding timelines. FINDING: The imagery community is not currently able to satisfy the requirements for both immediate and detailed analysis. These new tools will encompass a broad range of capabilities. In the interim, the emphasis should be on providing tools that will greatly speed up the analysts' ability to access and integrate information. Analysts need softcopy workstations that allow for timely retrieval of current and archived imagery with no degradation in quality. Softcopy exploitation will result in significant efficiencies. It will streamline the dissemination, storage and retrieval of imagery and will enhance the ability of analysts to exploit the full range of available data. It will facilitate the integration of classified, commercial and theater imagery, and will allow analysts to quickly acquire the "best" images of a target (assuming required selection algorithms are developed). The ability to perform mensuration from imagery obtained from multiple sensors at a single workstation will be a significant enhancement. RECOMMENDATION: The IC should move aggressively to infuse new technologies into the IMC, such as automatic target recognition capabilities, in order to help streamline the exploitation process. Softcopy search of large land areas is a critical necessity yet this tool remains extremely difficult to implement. Hence, currently, it must still be done on a light table. Softcopy search tools must be developed to enable efficient search of large amounts of data. Sufficient funding in R&D in this area must be accommodated. For the future, the best knowledge-based tools should be made available: on-line access to integrated databases from the analysts' desktops; numerous data sources available on-line (maps and intell reporting) at different security levels; simplified product lines in a limited number of formats; and the ability to receive requests on-line and distribute responses that way. A major investment is required to allow analysts to query, browse and exploit from large, digital image product libraries which use supercomputing and massive data storage technology. Providing this kind of access could greatly increase the amount of time an analyst spends on analysis. Direct interface of imagery with global geospatial information based on a standard coordinate system is required. Automated image examination technology must be pursued. Softcopy exploitation will be the norm; yet softcopy search will require high-speed computing, data storage and management capabilities in the gigaflop range of speed. Tools are needed to accomplish tonal dynamic range manipulation and sharpening, geometric processing for warping or imagery perspective manipulation, and registering images to maps. Data compression, management and display technologies are needed simultaneously. Adaptive image compression schemes will be needed to allow imagery analysts to quickly assimilate information without waiting for the full-resolution image. Greater screen brightness and higher resolution are needed for search. Flat screens with great resolution are needed for tactical situations. Three-dimensional technology will be important (e.g., autostereoscopic, holographic, and lenticular) but screen displays will be needed that do not require special viewing goggles. As imagery analysts search, locate, ID and analyze pertinent imagery, the results will be documented in real-time upon a registered, geographic, information- based, vectored layer. Analytic and presentation aids such as map overlays, terrain displays and 3-D perspectives will be routine. We must capitalize on commonalities among digital imagery and mapping technologies. Superimposition techniques on up-to-date baseline images, maps, and graphics will be able to show changes in force and target dispositions. Such symbolic information overlaid on baseline displays could provide tactical users readily accessible information in a format required for his command and control function. Hardcopy to softcopy conversion must also be a priority due to the vast quantities of historical documents containing text, graphics and pictures that are stored in paper and film form. Conversion technologies are needed that provide basic indices automatically, preserve formats, and permit full text searches. One area that remains quite controversial is automated target recognition (ATR) systems. There are many analysts who view ATR systems as direct competition with their jobs. Then, there are others who doubt whether these systems will ever be able to replace the imagery analyst. We have taken a moderate approach to ATR. As stated earlier, we believe that a human will be required in the loop, at least for the next 10-15 years. At that time, it may be possible that technology will have advanced far enough to allow cognitive aspects (i.e., assessing meaning, separating significant from irrelevant data, integrating all available data to form analytical context, making sense of imagery-derived data in the current situation, and judging the significance of the findings) of the exploitation process to be performed by computers. In the interim, we need technology to help analysts be more efficient, not to replace them. Thus, because ATR and artificial intelligence (AI) are a long way from performing these cognitive functions, we recommend increased attention to assisted target recognition (ASTR) systems while continuing low level exploration of ATR systems. R&D must be focused and pursued diligently in these areas for both imaging and spectroradiometric sensing, as ASTR/ATR offer the only major advancement in imagery analysis productivity on the horizon. ASTR/ATR have the potential to help resolve one of the IMC's biggest problems. In recent years, imagery analysts have been forced to be selective in the imagery they exploit. With the amount of imagery collected increasingly greatly in the near future, this priority-based exploitation will be the norm. The remainder of the imagery will be "binned" into libraries for ready access, if needed at a later date. If no one looks at this imagery at all, nothing will be found. Thus, if assisted "alerting mechanisms" can be developed with low enough false alarm rates to search this excess imagery, then the efficiency of our human analysts is greatly enhanced. There are algorithms of significant value available today that could be used as alert mechanisms. For the future, reliable, totally automated aids to help filter large volumes of data and accurately cue imagery analysts to likely points of intelligence interest will be essential. We should look to architect a system where tasks are efficiently divided between people and machines, parceling out to each the jobs that they do best. Some tasks for computers might be to screen non-literal imagery so an imagery analyst does not have to look at it (as mentioned above). Total automation will depend on what kind of false alarm rate can be tolerated. This will depend on the mission to be supported. Hence, algorithms need to be very specific to the job. We should take the ATR problem and break it up into bins, depending on the problem we are trying to solve. Then we should consolidate the bins and ask ourselves what the value is of doing this automatically. An assessment of that value should be traded against the cost. Computers are persistent but not very cognitive. They can be very good at search, can find bright spots, can look at certain parts of the spectrum, etc. On the other hand, because computers are much better at certain jobs than are people, in the near-term, we should concentrate on those areas where computers outperform humans and perhaps aim for 50-70 percent automation over the next 10 years. For other processes, we should proceed at a much slower rate and aim for 10-20 percent automation. Early success in automated aids are more likely to occur in filtering large volumes of imagery data to the analysts. High performance image screening and semi-automatic image region cueing also show promise. For the future, ATR needs to move to context-based recognition, not just for single objects for single vehicles, but for units in the field and activity types within fixed facilities. We also need to look at automating exploitation of moving target indicator imagery. If ATR algorithms can be developed that provide a very high level of confidence, then perhaps this processing can be transferred to the collector to allow screening before the data is downlinked. Some enabling technologies that should be investigated include domain mediators (which will help to quickly modify ATR algorithms to different but similar targets) and knowledge engineering tools (automating identification cues, context cues). RECOMMENDATION: Aggressively pursue ASTR/ATR algorithm development, concentrating in the near-term on those areas where computers outperform humans. Technology integration for exploitation has not progressed further or more rapidly in the IMC primarily because there exists no single focal point within the imagery community with sufficient influence to foster change. Funding constraints have forced the IMC to focus on procuring only a small part of the full array of needed technologies. No exploitation R&D roadmap exists and different programs seek different technological solutions to similar needs in dissemination, exploitation databases and softcopy. Establishing a funding line specifically for exploitation system development and supporting R&D would assist greatly in addressing exploitation shortfalls. Requiring that such a funding line be tied to each new collection system would ensure adequate "downstream" resources are addressed. Required critical technologies that surfaced during interviews include softcopy exploitation, automated or assisted exploitation, spectral phenomenology, imagery training, multimedia reporting and information infrastructure, surge retrieval visualization and synthesis schools, automated downgrading declassification, and hardcopy-to-softcopy conversion. Exploitation systems must evolve to acquisition timelines of months not years to keep pace with technology changes. For acquisition, we have to accept a 90 or 95 percent solution and not hold out for 100 percent if a commercial capability is available. Recapitalization is another area of concern. What is the optimum recapitalization timeline when what you take off the shelf is obsolete in a year? Other areas that need to be pursued include efficient means of data entry (like transferring reports to INTELINK) and the capability to precisely align or "fuse" two or more images of the same target but which have been collected from different attitudes, sensors and/or platforms. Newer imagery types (such as multispectral sensing) are harder in terms of their type and the tasks that have to be done for exploitation. A large amount of technology is being pursued piecemeal in this area but there has been no real high priority given to go perform missions in these areas. In the R&D community, we spend an inordinate amount of funds and time constructing databases for testing algorithms. We need a Community, common, controlled test data set and Community standards on metrics so new algorithms can be measured against each other from a common baseline. This would allow for a quick and smooth transition to the analysts' work environment. The analysts' workstations must be flexible and user friendly. Connectivity via email, at a minimum, with the ability to work on a common white board via personal videoteleconferencing at the individual workstations as a goal, must be implemented among all imagery analysts, both national and military. The IMC should define the standards for imagery exploitation, yet allow decentralized execution. Thus, while the all-source imagery analyst of the future will need more inherent analytic capability than is required today, perhaps the tactical imagery analysts will not if they can correspond real-time with other analysts in a coherent manner. In essence, we must strive toward the "virtual" imagery community. (We would also venture that all analysts, not just imagery analysts, have access to this connectivity, thereby creating a "virtual" IC.) Accordingly, analysts must have user- selectable and filterable theater/national SIGINT-IMINT-HUMINT cross-database query, cueing, and collection request capabilities to facilitate the targeting process and other near-real-time (NRT) requirements. From an IMINT perspective, central digital imagery libraries will be needed and an inventory of available theater imagery should also be accessible on-line. A network of accessible distributed databases integrated with the existing national database should be created. This comprehensive database should have capabilities beyond the current target-oriented systems and allow both imagery analysts and customers to access different levels of information to meet specific needs. In the battlefield of the future, fulfilling those NRT collection, exploitation, and dissemination needs will be critical. Ensuring our timelines are faster than those of our adversaries, especially when those adversaries will themselves have access to military grade imagery, will require implementation of all of these recommendations. These issues must be addressed within the immediate future in order for the imagery workforce to be adequate in the next century. Though some competitive analysis is healthy, the majority of today's isolated and/or redundant imagery production occurs because we are unable to share data, analysis and products between sites. Security measures that guard against unauthorized accesses, both intentional and inadvertent, without stifling system performance, are also required. RECOMMENDATION: The IC must move to all-digital exploitation of imagery, with access to cross-INT databases. Move to a "virtual" analytic environment, i.e., one in which analysts are connected electronically. Increase funding to accelerate the procurement of softcopy (digital) workstations for imagery analysts. The product of the future will be one of merged data from every -INT. They will become less and less textual and more graphical. Geospatially referenced graphical reporting with standardized symbology will become the norm. This will also provide an acceptable method to help protect sources and capabilities. However, the customer must work with these analysts closely before times of crises so that the customer will trust that the symbols are accurate. In this way, perhaps, we can reduce the number of customers who feel they need the raw image, when in fact, all they really need is the imagery-derived data. This issue, though, may become a moot point, if the "virtual" connectivity discussed above becomes reality. If the new IMC infrastructure is done correctly, users will be able to pull the raw image if he needs it or pull the imagery derived information, all the while retaining email/videoteleconferencing connectivity with analysts within the community. Our perception of CIO's archival plan is that it does not include the raw imagery. This is a mistake. All information should be accessible. If this occurs, the biggest issue will be ensuring that the user who pulls the raw image also takes advantage of the imagery derived data. A common misconception is that the significant intelligence contained in an image is readily apparent to the average observer. While it is true that a consumer, using an identification key, could find on electro-optical imagery an SA-2 site because of its distinctive pattern, the user would not be able to tell if the site were real, dummy, or decoy. Imagery analysis has come a long way from the days of photointerpretation. A comprehensive, analytical, multisource approach to imagery exploitation is now the standard within the National Photographic Interpretation Center (NPIC) and the Defense Intelligence Agency (DIA), though generally not at the force application levels of the military. The IMC must be able to serve both types of customers (force planners and force application end users) and provide support in both types of situations -- where the immediate transmission of raw imagery is enough and where imagery derived information is essential. The "virtual" connectivity mentioned earlier will erase the need to limit the number of raw images required by the user, rendering this contentious issue irrelevant. Procurement of information processing equipment is, and will continue to be, an incredible challenge for an acquisition system built for the Industrial Age. Trillions of dollars are being spent by industry on information technologies. New products are coming out every six months with new generations of products being produced every 18 months. Our information processing needs cannot survive an acquisition system that takes five to 10 years to field new systems (6.1, 6.2, 6.3 type funding is unacceptable for information processing systems -- it mandates a long development cycle). We need to modernize our procedures to take advantage of current technology. Our adversaries certainly will. Along these lines, we need to take advantage of commercial advancements and determine whether a commercial product that fulfills 90 percent of our requirements is adequate compared to the cost to customize that product for the extra 10 percent. We need to make maximum use of commercial off-the-shelf (COTS) products which requires someone to inform, encourage, influence and pay vendors to encompass our specialized needs in their technology advancement efforts. Standards are also required so the "guy in the foxhole" can receive imagery data, but government standards need to follow commercial standards if we are to truly benefit from COTS products. A government-commercial bridge is required, and luckily, one already exists. The National Technology Alliance (NTA) with the National Information Display Laboratory and the National Media Laboratory is that bridge and should be encouraged and expanded. The NTA attempts (and succeeds) in influencing commercial capabilities to encompass government requirements. It provides one set of government requirements that commercial companies can deal with and provides the commercial standards back to the government to influence government decisions. We must practice ways to influence COTS systems before they come to the marketplace so they will be useful to the government. The NTA has been instrumental in saving several government programs while simultaneously influencing commercial standards to better support government requirements. They should be a mandatory participant in any new acquisition of information processing equipment. They should be given the legislative and budgetary freedom to field ACTD-type experiments until commercial companies can pick up the support. The Department of Defense (DoD) might benefit in non-intelligence matters from a similar alliance to help accelerate the fielding of commercial systems. RECOMMENDATION: Expand the purview of the National Technology Alliance (NTA), increasing its resources and flexibility to provide more rapid fielding of new technologies, and to exploit commercially available technology. One approach to setting up the imagery processing (data storage, retrieval, etc.) and communications infrastructure, which merits closer scrutiny, is to hire a systems integrator to run this process. Systems integrators (SI) can cut across organizational boundaries (when given that authority) and have the flexibility to recapitalize quickly in areas where technology turns over frequently. These SIs are commercial companies that provide this type of service for a broad array of users. Their ability to consolidate, delete duplication, quickly upgrade capability and reduce costs provides a model the IMC community should strive to achieve. Classification One of the biggest controversies today is the sharing of imagery with our allies in the Balkans. Intelligence data sharing will continue to dominate foreign relations issues for many years. Every day we hear about a new request in ever more divergent areas: environmental, law enforcement, disaster relief, etc. Questions arise: How do we provide the same level of battlefield knowledge to our allies and coalition partners, how do we provide information on disasters, how do we provide data to support U.S. policy decision, all while continuing to protect sources and methods? During the majority of our panels, the customer reiterated that in most cases, he does not require the raw image, only the imagery-derived information. These consumers can be served with graphical overlays which provide the imagery derived information without giving away technical capability. This has worked very well in the support NPIC gives to FEMA (Federal Emergency Management Agency). FEMA provides the LANDSAT or SPOT image and the NPIC analysts overlay those images with a graphical representation using standardized symbology. It is a very efficient process. However, again, in order for the customer to trust the information provided in these graphical overlays, he must train with them. Of course, in the 21st century, anyone will be able to buy either military grade imagery (one meter) commercially or the actual satellite itself as a turn-key system. Yet, again, we should look to graphical overlays and imagery derived information as the medium we use to share data. We should protect the billions of dollars we invest in these capabilities for as long as we can; once the capability is known, adversaries will undertake countermeasures to defeat/degrade its collection capabilities. In the interim, graphical overlays will have to suffice. We should also move to protect any future technology breakthroughs. Are we no longer concerned with maintaining a U.S.-only capability and protecting our investments? We need to put back into the psyche of the community that secrecy is a requirement, not an option, especially before we invest dollars in next generation systems. We must move to new collection that is not understood by our adversaries. Along these lines, we should move to develop dissemination systems that can handle multiple levels of classification. Asynchronous Transfer Mode (ATM) technology will allow numerous levels of classification to be passed over the same communications lines. We need to develop the capability to have multiple levels of information accessible from the same workstation. Dissemination Dissemination of intelligence information was touted as the biggest failure of the IC during DESERT STORM. Though it remains a challenge today, much has been done at the national level to define interfaces and standards. Communications will be discussed in another IC21 study, but the bottom line for today is that imagery data can be disseminated to the theater in a timely manner. Below theater is where the problems lie and no national organization is going to be able to fix it. DoD must take the challenge and mandate that each theater's unique mix of national, commercial, and theater imagery needs and systems conform to common dissemination standards and interfaces. FINDING: Imagery dissemination to the military below the Joint Task Force level remains a problem. CIO's A3I (Accelerated Architecture Acquisition Initiative) is the right vision: virtual imagery archives accessible at every level. However, here is a program that would benefit tremendously from a Systems Integrator (SI). As stated earlier, these are commercial SIs who have streamlined and reduced overhead for numerous commercial and government ventures, providing "infrastructure" type functions for an overall cost savings. A3I must establish a virtual imagery archive for all digital imagery and imagery products that is easy for users to access. Users will "pull" whatever imagery and products they require. It is in essence, the imagery component of total battle space information to the warrior as envisioned in the C4I concept. Yet, it is really just data storage, archiving and retrieval, and the future we envision will have virtual databases for data from all of the -INTs. Thus, instead of setting up another stovepiped system, we must ensure that A3I will be compatible in the future with a virtual multi-INT data retrieval and archival system. We are not convinced that this is what is occurring and, in fact, A3I has been downgraded because of inadequate funding resources. Also, the military has been very skeptical of A3I because it does not address improvements to the communications network below the CINC level. Though this is not the imagery community's responsibility, an SI might be better equipped to cross organizational lines to implement the infrastructure to support everyone's requirements. In the near term, though, A3I should not be criticized for things out of its purview. An SI might be able to ensure that the communications community is looking at A3I to provide the necessary bandwidth and that, with the advent of global broadcast and direct broadcast service, connectivity via these systems will be easily and quickly incorporated. Denial and Deception RECOMMENDATION: The IC must continue to examine and to field means by which to overcome "denial and deception" activities. Commercial Systems Commercial systems should be viewed as an adjunct to our national collectors. There are some who believe that the small satellite initiatives and declassification of national imagery will put the commercial companies out of business. However, the commercial imagery companies developed their systems with the aircraft imagery market as their main consumer, not just for the US government (USG). Our prediction is that commercial imagery will be just as important to the USG tomorrow as it is today. It will be a valuable augmenter of the national/tactical systems and the multispectral sensing will provide unique data. One area that should be pursued is whether the commercial systems can provide a "surge" capability that would allow more real time collection/receipt of imagery during a crisis (similar to US Air Force's current ownership of a SPOT collection terminal within the Balkan theater). One sorely needed improvement is a new process for USG users to procure commercial data. The current process takes months, using the Defense Mapping Agency (DMA) as the middleman, and the customer forced to bring his own money. We envision, as part of our imagery organization concept, a central point which would procure commercial imagery as required from a central pot of funding, authorized and appropriated for this purpose. These purchases would be made on behalf of the USG so that anyone within the USG could use the imagery. This imagery would be archived within the main national imagery library where any user could access it. The imagery organization would maintain the index of what imagery had been procured. Of course, the disadvantage to this is that the imagery organization could become the bottleneck for these purchases, pushing the customers to go out and make their own agreements with the commercial companies. This should be allowed as long as the imagery purchased gets incorporated into the national, not just the regional, library, that a consolidated list of imagery purchased is passed to the central repository for indexing, and if a common USG license is issued. This allows flexibility across the board. FINDING: The IC can use commercial imagery more effectively to meet some requirements. There are some proposals being considered within the IC that would encourage and allow our allies to buy a medium resolution version of our imagery satellite system. These systems would be exempt from the current "shutter control" mandated by Presidential Decision Directive (PDD-23). The rationale for this proposal stems from a concern about the US being able to maintain its lead in this technology area because of reduced USG funding. Through these sales, we would have more funding available to invest in future systems while getting increased coverage from these additional systems. This proposal seems to contradict itself; on the one hand, promoting commercial systems is a priority while on the other hand, it advocates building a USG system for foreign military sales (FMS) that would directly compete with those same commercial systems. We are also concerned about giving away our technological advantage in this area. We believe that the shutter control policy is a necessity today. However, we must assume that eventually systems will be proliferated with no such encumbrances and should look to reassess the policy at that time. We also believe that our WFOV small satellite program will not compete with commercial programs or give any more unfair advantage to one program over another. The four licensed programs have all made the decision to go ahead and develop these systems without government funding. Further, the commercial systems would be complementary. By applying adequate collection management, offloading requirements to the commercial systems is a smart move on our part. This would free up our systems to collect other priorities. The biggest difference between our WFOV and the one discussed earlier is that ours would not be made available for government-to-government sales. We would encourage sales of available commercial systems. RECOMMENDATION: The IC must improve its acquisition and use of commercially available imagery. Such imagery can be used in lieu of more costly national assets. As demands to share imagery with non-allies during multilateral operations increases, the use of commercial imagery is especially important to obviate security concerns. RECOMMENDATION: Set up an account for the easy purchase of commercial imagery, done under the common U.S. government licenses. A central repository and indexing system should be created for easy access by all users.
Back to Table of Contents
Page #IC21006 June 5, 1996