This report develops a random field model that describes gas plumes in LWIR remote sensing images. The random field model serves as a prior distribution that can be combined with LWIR data to produce a posterior that determines the probability that a gas plume exists in the scene and also maps the most probable location of any plume. The random field model is intended to work with a single pixel regression estimator--a regression model that estimates gas concentration on an individual pixel basis.
Revised: May 30, 2014 |
Published: September 30, 2008
Citation
Heasler P.G. 2008.Development of a Random Field Model for Gas Plume Detection in Multiple LWIR Images. PNNL-17873. Richland, WA: Pacific Northwest National Laboratory.