Heat Pipe Preheater for Low NOx Catalytic Combustor

Not Available In this project, we will develop new algorithms for invariant subpixel material identification in thermal infrared hyperspectral images obtained from space-based sensors. Systems based on these algorithms have the potential to achieve high probabilities of detection in combination with low false alarm rates for the identification of small low contrast targets in the presence of camouflage, concealment, and deception. The algorithms will exploit a target subspace model that is based on intrinsic material properties and infrared phenomenology as well as a background model that is estimated from the data. These models will be used as the foundation of statistical tests for identifying subpixel instances of target materials. Since the approach is based on physical models, the algorithms can be configured according to ranges of thermal conditions and scene geometries, target/background spectral properties, and sensor characteristics. This will allow us to determine fundamental limits on performance. We will also examine the accuracy of the models and the performance of the algorithms over a range of available thermal spectral data.