SBIR-STTR Award

Air Force needs relatively low cost, real-time, and automated systems to continuously track, tag, and locate (TTL) dismounts and vehicles by employing multiple layers of surveillance and reconnaissance sensors operating in tandem at macro and close-in levels. In this proposal Gitam Technologies, Inc. (GTI) in collaboration with Prof. John Kerekes at Rochester Institute Technology (RIT) propose coordinated and queued sensing using Airborne and ground-based hyperspectral, EO and IR sensors for 24/7 monitoring and detection of ground objects. GTI's hyperspectral detect/ID/Track expertise joins forces with RIT researchers and leverage their ongoing hyperspectral and EO/IR airborne sensing, vehicle tracking and sensor modeling research. Both archived imagery and possible new collections will provide data for algorithm development and testing under this proposed research. In addition, DIRSIG simulated imagery will be used to provide test data with precisely known truth. The real and synthetic data sets include urban scenarios with specific vehicles in known truth locations and their movement tracked between images. Complementary high resolution EO imagery will be used in conjunction with lower resolution hyperspectral imagery to develop and test cueing concepts. Empirical analyses will also be further extended through use of the FASSP analytical modeling tool.

Keywords:
Dismount And Vehicle Detection And Recognition, Hyperspectral Image Processing, Multispectal, Eo/Ir , Multiple Target Tracking, Automatic Target Recog

Recent conflicts and unconventional threats in urban regions present significant challenges. There is an urgent need for unambiguous, precise identification before precision and guided ammunition can be deployed with high confidence by the warfighters. The required robust targeting needs to be cued from and coordinated by a collection of orthogonal sensor modalities. The key to many sensor challenges lies in higher levels of dimensionality in the signature space, achieved by fusing the outputs of EO/HSI/Ladar/Video and RF sensors in multiple spectral bands, ESM intercepts, ground sensors, and other specialized devices. This is both more effective and more affordable than attempts to build single sensors with unaided ability to prosecute difficult targets. During Phase II, the GTI-RIT team will join forces with WSU researchers to leverage their expertise in hyperspectral/EO/radar/HRR detect/ID/Tracking of vehicles and dismounts. In addition to archived HYDICE imagery, new data collections involving multitude of sensors will provide data for algorithm development and testing under this proposed research. In addition, DIRSIG simulated imagery will be used to provide test data with precisely known truth. The real and synthetic data sets will include urban scenarios with specific vehicles in known truth locations and their movement tracked between images.

Keywords:
Hyperspectral Image Processing, Multispectal, Eo/Video/Ir , Dismount And Vehicle Detection And Recognition, Multiple Target Tracking, Automatic Target