SBIR-STTR Award

We will develop the SenseMod system for the modeling of HSI sensors. The new sensor models will be used to advance the capability of hyperspectral pre-processing algorithms. SenseMod is based on an innovative approach for sensor characterization and for representing spectral differences between sensor measurements and model-based signature predictions. These differences depend on calibration uncertainty, approximation error, and sensor noise. The sensor models will be incorporated into state-of-the-art algorithms for detection and characterization. The use of the new models will be particularly advantageous for the detection and characterization of targets with low spectral contrast. The new models will also allow for the utilization of optimized algorithms by sensor systems with a wide range of properties. We will demonstrate the ability of SenseMod to improve the performance of algorithms during Phase I using a range of HSI data acquired by several sensors. The algorithms will be structured to allow real-time operation on COTS hardware. A detailed commercialization plan is given for the new software.

Keywords:
Hyperspectral, Remote Sensing, Target Detection, Target Characterization, Pre-Processing Algorithms, Real-Time, Clutter Suppression, Model-Based

High-fidelity modeling tools can be used to generate signature predictions for the pre-processing of hyperspectral image data.  The accuracy of signature predictions depends on factors that include sensor calibration error, model approximation error, and sensor noise.  In this project we will develop the SenseMod software product for the recovery of hyperspectral sensor models and for the use of these models in pre-processing algorithms. SenseMod will be based on MODTRAN(TM) forward models, innovative computational methods, and new statistical models.  SenseMod will consider the visible through the short-wavelength infrared and the long-wavelength infrared spectral regions.  The new sensor models will be combined with advanced algorithms to allow the detection and characterization of low-spectral-contrast targets.  The sensor model recovery and pre-processing algorithms will be designed to run in real-time on COTS hardware.  We will assess SenseMod using large sets of imagery acquired by several different hyperspectral sensors.  We will also consider the transition of SenseMod for a variety of applications.

Benefit:
Due to its unique position in the marketplace, SenseMod has great potential for the DoD and Intelligence Community.  The new product will greatly improve the operational utility of airborne and space-based hyperspectral sensors for a range of day/night surveillance, reconnaissance, and targeting applications.  SenseMod also has a high likelihood of success for Commercial/Civil applications in areas such as cartography, forestry, perimeter monitoring, and agricultural and environmental monitoring.  HyperTech is already involved in several programs that will lead to commercialization opportunities for SenseMod.

Keywords:
Hyperspectral, Remote Sensing, Target Detection, Target Characterization, Pre-Processing Algorithms, Real-Time, Clutter Suppression, Model-Based