A Hyperspectral Gas Analysis System (HyGAS)
Abstract
A prototype software hyperspectral gas analysis system (HyGAS) for standoff (remote) gas detection, identification, and analysis was designed for use with developing hyperspectral imaging spectrometer systems. HyGAS incorporates traditional (FFT, background removal and library search) and innovative (neural nets for gas identification and linear discriminants for gas detection) processing methods. Neural net and linear discriminant techniques are uniquely suited for low signal-to-noise ratio field applications, processing data quickly, and reducing data storage requirements, all important considerations for real-time exploitation and automated applications. HyGAS incorporates an interactive display-based system, where five spatial images and fifteen spectral plots can be displayed simultaneously. HyGAS applies spectral techniques of gas analysis to hyperspectral image cubes, providing a method of detecting and identifying gases and mapping regional extent and concentration. HyGAS is supplied with synthetic image cubes to simulate datasets collected by a hyperspectral imaging system over a possible gas release event covering the LWIR spectral region ( approx. 8-12 microns) with specifications selected to match the point spectronometers used to develop the neural net and linear discriminant techniques.
Document Details
- Document Type
- Technical Report
- Publication Date
- Dec 01, 1997
- Accession Number
- ADA335801
Entities
People
- Jessica Sunshine
- Juliana Lo
- Pam Owensby
- Stan Zisk