Hyperspectral Target Signature and Background Clutter Modeling in Tree Regions

"In this project we will develop a system for modeling hyperspectraltarget signatures and background clutter in tree regions. The systemwill generate VNIR/SWIR and LWIR target signature predictions bycombining models for 3-D tree structure, leaf optical properties,3-D target geometry and physical properties, the atmospheric andthermal conditions, and sensor properties. The models will includethe important special cases of FOPEN targets and target/canopy spectralmixtures. We will also develop physics-based spectral/spatial modelsfor background clutter in tree regions. These models can be used forautomated tree region segmentation and for the estimation of tree andleaf parameters from image data. The accuracy of the new modelingtechniques will be evaluated by comparing the generated models with largesets of hyperspectral image data. We will also develop innovativealgorithms that use the new target signature and background cluttermodels for invariant target detection."