Multiple spectral spatial classification approach for hyperspectral data

Yuliya Tarabalka; Jón Atli Benediktsson; Jocelyn Chanussot; James C. Tilton

doi:10.1109/TGRS.2010.2062526

Multiple spectral spatial classification approach for hyperspectral data

Yuliya Tarabalka^*, Jón Atli Benediktsson, Jocelyn Chanussot, James C. Tilton

^*Corresponding author for this work

University of Iceland

Research output: Contribution to journal › Article › peer-review

214 Citations (Scopus)

Abstract

A new multiple-classifier approach for spectralspatial classification of hyperspectral images is proposed. Several classifiers are used independently to classify an image. For every pixel, if all the classifiers have assigned this pixel to the same class, the pixel is kept as a marker, i.e., a seed of the spatial region with a corresponding class label. We propose to use spectralspatial classifiers at the preliminary step of the marker-selection procedure, each of them combining the results of a pixelwise classification and a segmentation map. Different segmentation methods based on dissimilar principles lead to different classification results. Furthermore, a minimum spanning forest is built, where each tree is rooted on a classification-driven marker and forms a region in the spectralspatial classification map. Experimental results are presented for two hyperspectral airborne images. The proposed method significantly improves classification accuracies when compared with previously proposed classification techniques.

Original language	English
Article number	5570985
Pages (from-to)	4122-4132
Number of pages	11
Journal	IEEE Transactions on Geoscience and Remote Sensing
Volume	48
Issue number	11
DOIs	https://doi.org/10.1109/TGRS.2010.2062526
Publication status	Published - Nov 2010

Bibliographical note

Funding Information:
Manuscript received December 30, 2009; revised May 11, 2010. Date of publication September 13, 2010; date of current version October 27, 2010. This work was supported in part by the Marie Curie Research Training Network “HYPER-I-NET.” Y. Tarabalka is with the Faculty of Electrical and Computer Engineering, University of Iceland, 107 Reykjavik, Iceland, and also with the Grenoble Images Speech Signals and Automatics Laboratory (GIPSA-Lab), Grenoble Institute of Technology (INPG), 38402 Saint-Martin-d’Hères Cedex, France (e-mail: yuliya.tarabalka@hyperinet.eu).

Other keywords

Classification
hyperspectral images
minimum spanning forest (MSF)
multiple classifiers (MCs)
segmentation

Access to Document

10.1109/TGRS.2010.2062526

Cite this

@article{f122f8622ecf439094e5e3cbd9a5d75d,

title = "Multiple spectral spatial classification approach for hyperspectral data",

abstract = "A new multiple-classifier approach for spectralspatial classification of hyperspectral images is proposed. Several classifiers are used independently to classify an image. For every pixel, if all the classifiers have assigned this pixel to the same class, the pixel is kept as a marker, i.e., a seed of the spatial region with a corresponding class label. We propose to use spectralspatial classifiers at the preliminary step of the marker-selection procedure, each of them combining the results of a pixelwise classification and a segmentation map. Different segmentation methods based on dissimilar principles lead to different classification results. Furthermore, a minimum spanning forest is built, where each tree is rooted on a classification-driven marker and forms a region in the spectralspatial classification map. Experimental results are presented for two hyperspectral airborne images. The proposed method significantly improves classification accuracies when compared with previously proposed classification techniques.",

keywords = "Classification, hyperspectral images, minimum spanning forest (MSF), multiple classifiers (MCs), segmentation",

author = "Yuliya Tarabalka and Benediktsson, {J{\'o}n Atli} and Jocelyn Chanussot and Tilton, {James C.}",

note = "Funding Information: Manuscript received December 30, 2009; revised May 11, 2010. Date of publication September 13, 2010; date of current version October 27, 2010. This work was supported in part by the Marie Curie Research Training Network “HYPER-I-NET.” Y. Tarabalka is with the Faculty of Electrical and Computer Engineering, University of Iceland, 107 Reykjavik, Iceland, and also with the Grenoble Images Speech Signals and Automatics Laboratory (GIPSA-Lab), Grenoble Institute of Technology (INPG), 38402 Saint-Martin-d{\textquoteright}H{\`e}res Cedex, France (e-mail: yuliya.tarabalka@hyperinet.eu).",

year = "2010",

month = nov,

doi = "10.1109/TGRS.2010.2062526",

language = "English",

volume = "48",

pages = "4122--4132",

journal = "IEEE Transactions on Geoscience and Remote Sensing",

issn = "0196-2892",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

number = "11",

}

TY - JOUR

T1 - Multiple spectral spatial classification approach for hyperspectral data

AU - Tarabalka, Yuliya

AU - Benediktsson, Jón Atli

AU - Chanussot, Jocelyn

AU - Tilton, James C.

N1 - Funding Information: Manuscript received December 30, 2009; revised May 11, 2010. Date of publication September 13, 2010; date of current version October 27, 2010. This work was supported in part by the Marie Curie Research Training Network “HYPER-I-NET.” Y. Tarabalka is with the Faculty of Electrical and Computer Engineering, University of Iceland, 107 Reykjavik, Iceland, and also with the Grenoble Images Speech Signals and Automatics Laboratory (GIPSA-Lab), Grenoble Institute of Technology (INPG), 38402 Saint-Martin-d’Hères Cedex, France (e-mail: yuliya.tarabalka@hyperinet.eu).

PY - 2010/11

Y1 - 2010/11

N2 - A new multiple-classifier approach for spectralspatial classification of hyperspectral images is proposed. Several classifiers are used independently to classify an image. For every pixel, if all the classifiers have assigned this pixel to the same class, the pixel is kept as a marker, i.e., a seed of the spatial region with a corresponding class label. We propose to use spectralspatial classifiers at the preliminary step of the marker-selection procedure, each of them combining the results of a pixelwise classification and a segmentation map. Different segmentation methods based on dissimilar principles lead to different classification results. Furthermore, a minimum spanning forest is built, where each tree is rooted on a classification-driven marker and forms a region in the spectralspatial classification map. Experimental results are presented for two hyperspectral airborne images. The proposed method significantly improves classification accuracies when compared with previously proposed classification techniques.

AB - A new multiple-classifier approach for spectralspatial classification of hyperspectral images is proposed. Several classifiers are used independently to classify an image. For every pixel, if all the classifiers have assigned this pixel to the same class, the pixel is kept as a marker, i.e., a seed of the spatial region with a corresponding class label. We propose to use spectralspatial classifiers at the preliminary step of the marker-selection procedure, each of them combining the results of a pixelwise classification and a segmentation map. Different segmentation methods based on dissimilar principles lead to different classification results. Furthermore, a minimum spanning forest is built, where each tree is rooted on a classification-driven marker and forms a region in the spectralspatial classification map. Experimental results are presented for two hyperspectral airborne images. The proposed method significantly improves classification accuracies when compared with previously proposed classification techniques.

KW - Classification

KW - hyperspectral images

KW - minimum spanning forest (MSF)

KW - multiple classifiers (MCs)

KW - segmentation

UR - http://www.scopus.com/inward/record.url?scp=78049241977&partnerID=8YFLogxK

U2 - 10.1109/TGRS.2010.2062526

DO - 10.1109/TGRS.2010.2062526

M3 - Article

AN - SCOPUS:78049241977

VL - 48

SP - 4122

EP - 4132

JO - IEEE Transactions on Geoscience and Remote Sensing

JF - IEEE Transactions on Geoscience and Remote Sensing

SN - 0196-2892

IS - 11

M1 - 5570985

ER -

Multiple spectral spatial classification approach for hyperspectral data

Abstract

Bibliographical note

Other keywords

Access to Document

Fingerprint

Cite this