Semisupervised local discriminant analysis for feature extraction in hyperspectral images

Wenzhi Liao, Aleksandra Pizurica, Paul Scheunders, Wilfried Philips, Youguo Pi

Research output: Contribution to journalArticle

111 Citations (Scopus)

Abstract

We propose a novel semisupervised local discriminant analysis method for feature extraction in hyperspectral remote sensing imagery, with improved performance in both ill-posed and poor-posed conditions. The proposed method combines unsupervised methods (local linear feature extraction methods and supervised method (linear discriminant analysis) in a novel framework without any free parameters. The underlying idea is to design an optimal projection matrix, which preserves the local neighborhood information inferred from unlabeled samples, while simultaneously maximizing the class discrimination of the data inferred from the labeled samples. Experimental results on four real hyperspectral images demonstrate that the proposed method compares favorably with conventional feature extraction methods.
LanguageEnglish
Pages184-198
Number of pages15
JournalIEEE Transactions on Geoscience and Remote Sensing
Volume51
Issue number1
DOIs
Publication statusPublished - 31 Jan 2013

Fingerprint

Discriminant analysis
discriminant analysis
Feature extraction
extraction method
Remote sensing
imagery
method
remote sensing
matrix

Keywords

  • feature extraction
  • hyperspectral remote sensing
  • semisupervised
  • classification

Cite this

Liao, Wenzhi ; Pizurica, Aleksandra ; Scheunders, Paul ; Philips, Wilfried ; Pi, Youguo. / Semisupervised local discriminant analysis for feature extraction in hyperspectral images. In: IEEE Transactions on Geoscience and Remote Sensing. 2013 ; Vol. 51, No. 1. pp. 184-198.
@article{7888addf1ad5461699c9dda7b75a8718,
title = "Semisupervised local discriminant analysis for feature extraction in hyperspectral images",
abstract = "We propose a novel semisupervised local discriminant analysis method for feature extraction in hyperspectral remote sensing imagery, with improved performance in both ill-posed and poor-posed conditions. The proposed method combines unsupervised methods (local linear feature extraction methods and supervised method (linear discriminant analysis) in a novel framework without any free parameters. The underlying idea is to design an optimal projection matrix, which preserves the local neighborhood information inferred from unlabeled samples, while simultaneously maximizing the class discrimination of the data inferred from the labeled samples. Experimental results on four real hyperspectral images demonstrate that the proposed method compares favorably with conventional feature extraction methods.",
keywords = "feature extraction, hyperspectral remote sensing, semisupervised, classification",
author = "Wenzhi Liao and Aleksandra Pizurica and Paul Scheunders and Wilfried Philips and Youguo Pi",
year = "2013",
month = "1",
day = "31",
doi = "10.1109/TGRS.2012.2200106",
language = "English",
volume = "51",
pages = "184--198",
journal = "IEEE Transactions on Geoscience and Remote Sensing",
issn = "0196-2892",
publisher = "IEEE",
number = "1",

}

Semisupervised local discriminant analysis for feature extraction in hyperspectral images. / Liao, Wenzhi; Pizurica, Aleksandra; Scheunders, Paul; Philips, Wilfried; Pi, Youguo.

In: IEEE Transactions on Geoscience and Remote Sensing, Vol. 51, No. 1, 31.01.2013, p. 184-198.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Semisupervised local discriminant analysis for feature extraction in hyperspectral images

AU - Liao, Wenzhi

AU - Pizurica, Aleksandra

AU - Scheunders, Paul

AU - Philips, Wilfried

AU - Pi, Youguo

PY - 2013/1/31

Y1 - 2013/1/31

N2 - We propose a novel semisupervised local discriminant analysis method for feature extraction in hyperspectral remote sensing imagery, with improved performance in both ill-posed and poor-posed conditions. The proposed method combines unsupervised methods (local linear feature extraction methods and supervised method (linear discriminant analysis) in a novel framework without any free parameters. The underlying idea is to design an optimal projection matrix, which preserves the local neighborhood information inferred from unlabeled samples, while simultaneously maximizing the class discrimination of the data inferred from the labeled samples. Experimental results on four real hyperspectral images demonstrate that the proposed method compares favorably with conventional feature extraction methods.

AB - We propose a novel semisupervised local discriminant analysis method for feature extraction in hyperspectral remote sensing imagery, with improved performance in both ill-posed and poor-posed conditions. The proposed method combines unsupervised methods (local linear feature extraction methods and supervised method (linear discriminant analysis) in a novel framework without any free parameters. The underlying idea is to design an optimal projection matrix, which preserves the local neighborhood information inferred from unlabeled samples, while simultaneously maximizing the class discrimination of the data inferred from the labeled samples. Experimental results on four real hyperspectral images demonstrate that the proposed method compares favorably with conventional feature extraction methods.

KW - feature extraction

KW - hyperspectral remote sensing

KW - semisupervised

KW - classification

UR - http://hdl.handle.net/1854/LU-2962890

U2 - 10.1109/TGRS.2012.2200106

DO - 10.1109/TGRS.2012.2200106

M3 - Article

VL - 51

SP - 184

EP - 198

JO - IEEE Transactions on Geoscience and Remote Sensing

T2 - IEEE Transactions on Geoscience and Remote Sensing

JF - IEEE Transactions on Geoscience and Remote Sensing

SN - 0196-2892

IS - 1

ER -