Spotter: Target Acquisition in Large Format Imagery
Small Business Information
21ST CENTURY TECHNOLOGIES, INC.
4515 Seton Center Parkway, Suite 320, Austin, TX, 78759
Senior Research Scientist
Senior Research Scientist
AbstractThe Spotter Phase 2 effort augments the successful Phase 1 approach to provide advanced automatic target acquisition algorithms tailored to exploit large format (LF) imagery. 21st Century Technologies, Inc. (21CT) proposes a suite of detection algorithms specifically designed to address the challenges of wide area aerial surveillance and reconnaissance. Spotter automatically detects vehicular targets of interest in LF imagery using two complementary target acquisition approaches: a novel rotationally-invariant extension to Haar wavelet features, and a phaseless analysis of regional frequency content inspired by research in human visual search. Spotter fuses acquisition results to build a georegistered target likelihood map to enable downstream exploitation. Successful LF exploitation yields superior operational performance and reduced analyst workload. Spotter combines the results of complementary detection approaches in a unique cascaded detection, aggregation, and fusion framework that builds consensus over all available inputs and detection algorithms to obtain the most accurate estimation of target presence. By doing so, Spotter will surpass the operational performance of existing approaches to wide area aerial imagery exploitation. Spotters capabilities can significantly increase imagery analyst accuracy and productivity, and can also provide the basis for an automated layered sensing system, further supporting the intelligence, surveillance, and reconnaissance mission. BENEFITS: Large format (LF) sensors are being developed to provide wide area coverage at unprecedented resolutions. Employed aboard UMS or other aerial platforms, these sensors can support a variety of intelligence, surveillance, and reconnaissance (ISR) missions. LF sensors present unique technical challenges, and their ever-increasing volume of imagery poses significant operational challenges to human analysts. Spotter automates the process of detecting vehicular targets in LF imagery, significantly reducing the volume of data presented to an analyst and increasing the operational value of the resulting intelligence product. LF imagery provides adequate resolution for trained experts to identify particular classes of vehicles, yet inadequate resolution for most current automatic detection algorithms. Spotter provides algorithms designed specifically for LF imagery which address the challenge of relatively low on-target resolution, as well as the need to efficiently process the large data volume. The detection performance of different approaches varies based on scene content and environmental factors. Spotter combines a suite of algorithms with a novel fusion strategy to provide robust detection performance in a wide variety of scenarios. Spotter can be applied in a variety of ways to provide military, non-military, and commercial markets the ability to locate and track vehicular targets in wide area aerial surveillance imagery. Military applications include improved image exploitation in surveillance, reconnaissance, and force protection missions. Non-military applications include highway surveillance and traffic monitoring. Additionally, Spotters geo-location of targets of interest provides the foundation for autonomous control of layered sensing systems, which is a revolutionary approach to the management of military sensor platforms. The structure of the Phase 2 effort maximizes the commercial attractiveness of our final capabilities. In addition to our core technology development, the effort provides compelling quantitative evaluations of the approachs superior performance using operationally relevant aerial surveillance imagery. Together, these provide a solid foundation for our efforts to commercialize and operationalize Spotter capabilities, and to secure Phase 3 funding.
* information listed above is at the time of submission.