Abstract
The regularized feature extraction methods for hyperspectral data classification were studied. The regularization algorithms worked for both parametric and nonparametric within-class scatter matrix. Real data experiment and simulated results show that the nonparametric weighted feature extraction (NWFE) is better method than the nonparametric discriminant analysis (NDA) and discriminant analysis feature extraction (DAFE).
Original language | English |
---|---|
Pages | 1767-1769 |
Number of pages | 3 |
State | Published - 24 Nov 2003 |
Event | 2003 IGARSS: Learning From Earth's Shapes and Colours - Toulouse, France Duration: 21 Jul 2003 → 25 Jul 2003 |
Conference
Conference | 2003 IGARSS: Learning From Earth's Shapes and Colours |
---|---|
Country | France |
City | Toulouse |
Period | 21/07/03 → 25/07/03 |
Keywords
- Feature extraction
- Hyperspectral data classification
- Regularization