27 August 2008 A family of distributions for the error term in linear mixing models for hyperspectral images
Author Affiliations +
Abstract
A traditional linear-mixing model with a structured background used in the hyperspectral imaging literature often assumes normality (Gaussianity) of the error term. This assumption is often questioned. In previous research, we show that the normal (Gaussian) distribution gives only a very crude approximation to the actual error term distribution. In this paper, we use a broader class of distributions called exponential power (or error) distributions. We investigate suitability of those distributions using a specific example of an AVIRIS hyperspectral image. We demonstrate that the exponential power distributions provide a satisfactory description of the marginal error term distributions for the AVIRIS hyperspectral image used in this paper.
© (2008) COPYRIGHT Society of Photo-Optical Instrumentation Engineers (SPIE). Downloading of the abstract is permitted for personal use only.
Peter Bajorski, Peter Bajorski, } "A family of distributions for the error term in linear mixing models for hyperspectral images", Proc. SPIE 7086, Imaging Spectrometry XIII, 70860E (27 August 2008); doi: 10.1117/12.793380; https://doi.org/10.1117/12.793380
PROCEEDINGS
8 PAGES


SHARE
Back to Top