Description:
OBJECTIVE: Investigate algorithms that use compressive sensing to improve processing, delivery, and storage of wide-area surveillance products such as radio frequency and wide-area motion imagery to save critical bandwidth and accelerate detection, recognition and tracking for tactical users. DESCRIPTION: Compressive sensing is a relatively new form of data sampling that shows promise to greatly reduce the amount of information that is required to acquire and reconstruct information from sources such as radio frequency, synthetic aperture radar, and electro-optical sensors. The theory of compressive sensing has some interesting practical applications in the processing and exploitation of images, signals and other structured data. In some cases it can be used as an improved form of compression and reconstruction of sensor data. In other cases it can be used to detect, classify and estimate with reduced dimensionality and thereby provide increased operational rates over the original sources. These particular applications of compressive sensing lend themselves well to address the various applications of wide-area persistent surveillance. Intelligence, Surveillance, and Reconnaissance sensors are now producing data that has greater density and wider field of regard. These sensors are being integrated into a mix of manned and unmanned platforms which will provide an increasing persistent stare over large areas. The volume and amount of dense media being produced is already beyond the capacity of analysts to review. As a result, automated exploitation is more important than ever. The dense sensor media sometimes cannot be reasonably distributed in raw form across the Global Information Grid (GIG). Innovative techniques are needed to accomplish this on the sensor and on the server which receives and distributes the exploitated data product. In order to address wide-area persistent surveillance in its operational form, the algorithms should be applied to some degree on the sensor platforms as well as the sensor server platforms across the GIG, and for automated exploitation in server farm concentrator points. For instance, in wide-area video, compressive sensing could speed up regional detections and stream high resolution areas of interest and lower resolution backgrounds for reference. Such applications can be applied to maritime operations where vessel detection could occur on shipboard sensors and for higher order ship classification at a ground station server. PHASE I: Research uses of compressive sensing algorithms to find the best application areas to speed up exploitation and distribution of media produced by wide-area persistent surveillance sensors. Develop small illustrative prototype implementations and experiments that prove the feasibility for improvements and the qualitative aspects of the proposed implementations. PHASE II: Design a prototype compressive sensing software application suite that can address processing on the sensor platform as well as on a DCGS-N server. The prototype should illustrate techniques that automatically detect areas of concern and reduce the need to send wide-area surveillance data across the GIG. The system should focus on compression, reconstruction, and detection and also assist higher level functions such as identification, tracking and behavioral clues. Prototype demonstration could be at the SECRET classification level. PHASE III: Transition the technology into the Distributed Common Ground Station - Navy system to allow it to handle wide-area surveillance products from unmanned aircraft systems and shipboard surveillance systems. PRIVATE SECTOR COMMERCIAL POTENTIAL/DUAL-USE APPLICATIONS: Homeland security and port security can all use wide-area surveillance applications.
