Statistics enhancement in hyperspectral data analysis using spectral-spatial labeling, the EM algorithm, and the Leave-One-Out covariance estimator

Proceedings of SPIE - The International Society for Optical Engineering

Volumn 3438, Issue , 1998, Pages 183-190

  Hsieh, Pi Fuei ^a   Landgrebe, David ^a  

^a PURDUE UNIVERSITY   (United States)

Author keywords

Classification; Hyperspectral analysis; Multispectral analysis; Parameter estimation; Small sample size problem; The EM algorithm

Indexed keywords

ALGORITHMS; CLASSIFICATION (OF INFORMATION); PARAMETER ESTIMATION; SPECTRUM ANALYSIS; STATISTICAL METHODS;

STATISTICS ENHANCEMENT;

DATA REDUCTION;

EID: 0032224562     PISSN: 0277786X     EISSN: None     Source Type: Conference Proceeding    
DOI: 10.1117/12.328101     Document Type: Conference Paper

References (9)

1. Analyzing high dimensional multispectral data
2. Maximum likelihood estimation from incomplete data via EM algorithm
3. The effect of unlabeled samples in reducing the small sample size problem and mitigating the Hughes phenomenon
4. Covariance matrix estimation and classification with limited training data
5. Classification of remote sensing images having high spectral resolution
6. Effects of sample size in classifier design
7. Projection pursuit in high dimensional data reduction: Initial conditions, feature selection and the assumption of normality