Feature extraction for hyperspectral images based on semi-supervised local linear discriminant analysis

Wenzhi Liao, Aleksandra Pizurica, Wilfried Philips, Youguo Pi, Uwe Stilla (Editor), P Gamba, C Juergens (Editor), D Maktav (Editor)

Research output: Contribution to conferencePaper

12 Citations (Scopus)

Abstract

We propose a novel semi-supervised local discriminant analysis (SELD) method for feature extraction in hyperspectral remote sensing imagery. The proposed method combines a supervised method (Linear Discriminant Analysis (LDA)) and an unsupervised method (Neighborhood Preserving Embedding (NPE)) without any free parameters. The underlying idea is to design optimal projection vectors, which can discover the global discriminant structure of the available labeled samples while preserving the local neighborhood spatial structure of the unlabeled samples. Furthermore, in our approach the number of extracted feature bands is no longer limited by the number of classes, which is a disadvantage of LDA. Experimental results demonstrate that the proposed method outperforms consistently other related semi-supervised methods and that it is also much more stable when the percentage of the labeled samples changes.
LanguageEnglish
Pages401-404
Number of pages4
DOIs
Publication statusPublished - 5 May 2011
EventJoint Urban Remote Sensing Event (JURSE - 2011) - Munich, Germany
Duration: 11 Apr 201113 Apr 2011

Conference

ConferenceJoint Urban Remote Sensing Event (JURSE - 2011)
Abbreviated titleJURSE 2011
CountryGermany
CityMunich
Period11/04/1113/04/11

Fingerprint

discriminant analysis
pattern recognition
preserving
imagery
embedding
remote sensing
projection
method

Keywords

  • feature extraction
  • hyperspectral remote sensing
  • classification
  • principal component analysis
  • geophysical image processing
  • image colour analysis

Cite this

Liao, W., Pizurica, A., Philips, W., Pi, Y., Stilla, U. (Ed.), Gamba, P., ... Maktav, D. (Ed.) (2011). Feature extraction for hyperspectral images based on semi-supervised local linear discriminant analysis. 401-404. Paper presented at Joint Urban Remote Sensing Event (JURSE - 2011), Munich, Germany. https://doi.org/10.1109/JURSE.2011.5764804
Liao, Wenzhi ; Pizurica, Aleksandra ; Philips, Wilfried ; Pi, Youguo ; Stilla, Uwe (Editor) ; Gamba, P ; Juergens, C (Editor) ; Maktav, D (Editor). / Feature extraction for hyperspectral images based on semi-supervised local linear discriminant analysis. Paper presented at Joint Urban Remote Sensing Event (JURSE - 2011), Munich, Germany.4 p.
@conference{6e53db08caf442c68c6c4c5365617522,
title = "Feature extraction for hyperspectral images based on semi-supervised local linear discriminant analysis",
abstract = "We propose a novel semi-supervised local discriminant analysis (SELD) method for feature extraction in hyperspectral remote sensing imagery. The proposed method combines a supervised method (Linear Discriminant Analysis (LDA)) and an unsupervised method (Neighborhood Preserving Embedding (NPE)) without any free parameters. The underlying idea is to design optimal projection vectors, which can discover the global discriminant structure of the available labeled samples while preserving the local neighborhood spatial structure of the unlabeled samples. Furthermore, in our approach the number of extracted feature bands is no longer limited by the number of classes, which is a disadvantage of LDA. Experimental results demonstrate that the proposed method outperforms consistently other related semi-supervised methods and that it is also much more stable when the percentage of the labeled samples changes.",
keywords = "feature extraction, hyperspectral remote sensing, classification, principal component analysis, geophysical image processing, image colour analysis",
author = "Wenzhi Liao and Aleksandra Pizurica and Wilfried Philips and Youguo Pi and Uwe Stilla and P Gamba and C Juergens and D Maktav",
year = "2011",
month = "5",
day = "5",
doi = "10.1109/JURSE.2011.5764804",
language = "English",
pages = "401--404",
note = "Joint Urban Remote Sensing Event (JURSE - 2011), JURSE 2011 ; Conference date: 11-04-2011 Through 13-04-2011",

}

Liao, W, Pizurica, A, Philips, W, Pi, Y, Stilla, U (ed.), Gamba, P, Juergens, C (ed.) & Maktav, D (ed.) 2011, 'Feature extraction for hyperspectral images based on semi-supervised local linear discriminant analysis' Paper presented at Joint Urban Remote Sensing Event (JURSE - 2011), Munich, Germany, 11/04/11 - 13/04/11, pp. 401-404. https://doi.org/10.1109/JURSE.2011.5764804

Feature extraction for hyperspectral images based on semi-supervised local linear discriminant analysis. / Liao, Wenzhi; Pizurica, Aleksandra; Philips, Wilfried; Pi, Youguo; Stilla, Uwe (Editor); Gamba, P; Juergens, C (Editor); Maktav, D (Editor).

2011. 401-404 Paper presented at Joint Urban Remote Sensing Event (JURSE - 2011), Munich, Germany.

Research output: Contribution to conferencePaper

TY - CONF

T1 - Feature extraction for hyperspectral images based on semi-supervised local linear discriminant analysis

AU - Liao, Wenzhi

AU - Pizurica, Aleksandra

AU - Philips, Wilfried

AU - Pi, Youguo

AU - Gamba, P

A2 - Stilla, Uwe

A2 - Juergens, C

A2 - Maktav, D

PY - 2011/5/5

Y1 - 2011/5/5

N2 - We propose a novel semi-supervised local discriminant analysis (SELD) method for feature extraction in hyperspectral remote sensing imagery. The proposed method combines a supervised method (Linear Discriminant Analysis (LDA)) and an unsupervised method (Neighborhood Preserving Embedding (NPE)) without any free parameters. The underlying idea is to design optimal projection vectors, which can discover the global discriminant structure of the available labeled samples while preserving the local neighborhood spatial structure of the unlabeled samples. Furthermore, in our approach the number of extracted feature bands is no longer limited by the number of classes, which is a disadvantage of LDA. Experimental results demonstrate that the proposed method outperforms consistently other related semi-supervised methods and that it is also much more stable when the percentage of the labeled samples changes.

AB - We propose a novel semi-supervised local discriminant analysis (SELD) method for feature extraction in hyperspectral remote sensing imagery. The proposed method combines a supervised method (Linear Discriminant Analysis (LDA)) and an unsupervised method (Neighborhood Preserving Embedding (NPE)) without any free parameters. The underlying idea is to design optimal projection vectors, which can discover the global discriminant structure of the available labeled samples while preserving the local neighborhood spatial structure of the unlabeled samples. Furthermore, in our approach the number of extracted feature bands is no longer limited by the number of classes, which is a disadvantage of LDA. Experimental results demonstrate that the proposed method outperforms consistently other related semi-supervised methods and that it is also much more stable when the percentage of the labeled samples changes.

KW - feature extraction

KW - hyperspectral remote sensing

KW - classification

KW - principal component analysis

KW - geophysical image processing

KW - image colour analysis

UR - http://hdl.handle.net/1854/LU-1216543

U2 - 10.1109/JURSE.2011.5764804

DO - 10.1109/JURSE.2011.5764804

M3 - Paper

SP - 401

EP - 404

ER -

Liao W, Pizurica A, Philips W, Pi Y, Stilla U, (ed.), Gamba P et al. Feature extraction for hyperspectral images based on semi-supervised local linear discriminant analysis. 2011. Paper presented at Joint Urban Remote Sensing Event (JURSE - 2011), Munich, Germany. https://doi.org/10.1109/JURSE.2011.5764804