An infinite Gaussian mixture model with its application in hyperspectral unmixing

作者：

Highlights：

• A new mixture model called IGMM is proposed to hyperspectral data.

• TSS is used to direct the trends of the number of Gaussian components.

• A Metropolis-within-Gibbs sampler is used for all the parameters.

摘要

•A new mixture model called IGMM is proposed to hyperspectral data.•TSS is used to direct the trends of the number of Gaussian components.•A Metropolis-within-Gibbs sampler is used for all the parameters.

论文关键词：Hyperspectral unmixing,Bayesian inference,Hyperspectral image,Infinite Gaussian mixture model,Metropolis-within-Gibbs

论文评审过程：Available online 12 October 2014.

论文官网地址：https://doi.org/10.1016/j.eswa.2014.09.059