Decision-Support and Feature Extraction with the Full Spectrum Retrieval (FUSR) Toolkit

摘要

A decision-support tool for risk management and global change monitoring based on remotely sensed spectral imagery data, called the Full Spectrum Retrieval Toolkit (FUSR) is described. Conventional spectral remote sensing technologies - such as atmospheric compensation, end-member identification, sub-pixel detection, and linear unmixing algorithms - are assimilated with an advanced, physics-based technique known as a model matching retrieval (MMR), supporting an operational capability that spans the optical domain (0.4 μm to 16 μm). Because the contributions of the atmosphere, surface, and sensor response are modeled and/or retrieved simultaneously, FUSR also represents a powerful and unique tool for performing comprehensive CONOPS and sensor design tradeoff studies. The underlying retrieval algorithm, known as a square-root information filter (SRIF), also provides a mathematical infrastructure for data integration supporting multi-sensor, multi-look observational scenarios. Data processing and feature extraction capabilities are demonstrated through an analysis of AVIRIS imagery collected over Cuprite, NV.