NRL conducts research to facilitate the use of hyperspectral imaging data. The data from hyperspectral imagers form very large data sets and may be supplemented by additional data in the form of ground truth, previous measurements, or data from other sensors such as synthetic aperture radar, lidars, thermal imagers, etc. Many physics-based algorithms have been developed over the years to gather particular information from the data. Examples would include algorithms for determining the chlorophyll content of deep water and/or other water column content. A class of algorithms also exist that are more general in nature and that generally analyze the data with no set type of output. The desired outputs include terrain categorization or anomaly detection. Examples of this type of algorithm would include principle components analysis, projection pursuit, and manifold coordinate analysis. A number of general approaches are being developed. One is PURSUIT (Projection Understanding and Recognition from a Suite of Intelligent Tools), a modular tool-set for statistical pattern recognition, suitable for a variety of applications. Another is the Optical Real-time Spectral Identification System (ORASIS). This system is based on convex set methods and has applications in many areas. More generally, the data exist in a high dimensional space. A series of algorithms is being developed that attempt to unravel the complicated nature of this type of data. One example is a new tool Manifold Coordinate Analysis Workshop (MCAW), which has been designed to derive intrinsic coordinate descriptions of high-dimensional data. The development of any analysis tool, whether dependent on physics-based modeling approaches or on clarifying the understanding of the structure of HSI and multisensor data, is appropriate.

 

key words

Hyperspectral; HSI; MSI; Lidar; SAR; Thermal imagers; Optics; Applied mathematics;

Citizenship:  Open to U.S. citizens and permanent residents

Level:  Open to Postdoctoral applicants