Global spatial and local spectral similarity-based group sparse representation for hyperspectral imagery classification

Haoyang Yu, Lianru Gao, Wenzhi Liao, Paolo Gamba, Bing Zhang

Research output: Contribution to conferencePaper

Abstract

Spectral-spatial classification has been widely exploited for hyperspectral imagery. However, current methods either focus on local spatial similarity or global nonlocal self-similarity (NLSS). In this paper, we propose novel methods to couple both global spatial similarity and local spectral similarity together in a single framework. In particular, our approaches exploit global spatial similarity by searching non-overlap nonlocal patches, whereas spectral similarity is determined locally within the found patches. Experimental results on two real hyperspectral data sets demonstrate the efficiency of the proposed methods, with 5%-7% (overall classification accuracy) improvements over approaches that only consider either global or local similarity.
Original languageEnglish
Pages3579-3582
Number of pages4
DOIs
Publication statusPublished - 5 Nov 2018
Event38th Annual IEEE International Geoscience and Remote Sensing Symposium, IGARSS 2018 - Valencia, Spain
Duration: 22 Jul 201827 Jul 2018

Conference

Conference38th Annual IEEE International Geoscience and Remote Sensing Symposium, IGARSS 2018
CountrySpain
CityValencia
Period22/07/1827/07/18

Fingerprint

imagery
method

Keywords

  • hyperspectral image
  • classification
  • nonlocal self-similarity
  • group sparse representation
  • support vector machines
  • spatial coherence
  • geophysical image processing

Cite this

Yu, H., Gao, L., Liao, W., Gamba, P., & Zhang, B. (2018). Global spatial and local spectral similarity-based group sparse representation for hyperspectral imagery classification. 3579-3582. Paper presented at 38th Annual IEEE International Geoscience and Remote Sensing Symposium, IGARSS 2018, Valencia, Spain. https://doi.org/10.1109/IGARSS.2018.8518227
Yu, Haoyang ; Gao, Lianru ; Liao, Wenzhi ; Gamba, Paolo ; Zhang, Bing. / Global spatial and local spectral similarity-based group sparse representation for hyperspectral imagery classification. Paper presented at 38th Annual IEEE International Geoscience and Remote Sensing Symposium, IGARSS 2018, Valencia, Spain.4 p.
@conference{419a77f6d809438f9a31fefd6fb37dd8,
title = "Global spatial and local spectral similarity-based group sparse representation for hyperspectral imagery classification",
abstract = "Spectral-spatial classification has been widely exploited for hyperspectral imagery. However, current methods either focus on local spatial similarity or global nonlocal self-similarity (NLSS). In this paper, we propose novel methods to couple both global spatial similarity and local spectral similarity together in a single framework. In particular, our approaches exploit global spatial similarity by searching non-overlap nonlocal patches, whereas spectral similarity is determined locally within the found patches. Experimental results on two real hyperspectral data sets demonstrate the efficiency of the proposed methods, with 5{\%}-7{\%} (overall classification accuracy) improvements over approaches that only consider either global or local similarity.",
keywords = "hyperspectral image, classification, nonlocal self-similarity, group sparse representation, support vector machines, spatial coherence, geophysical image processing",
author = "Haoyang Yu and Lianru Gao and Wenzhi Liao and Paolo Gamba and Bing Zhang",
year = "2018",
month = "11",
day = "5",
doi = "10.1109/IGARSS.2018.8518227",
language = "English",
pages = "3579--3582",
note = "38th Annual IEEE International Geoscience and Remote Sensing Symposium, IGARSS 2018 ; Conference date: 22-07-2018 Through 27-07-2018",

}

Yu, H, Gao, L, Liao, W, Gamba, P & Zhang, B 2018, 'Global spatial and local spectral similarity-based group sparse representation for hyperspectral imagery classification' Paper presented at 38th Annual IEEE International Geoscience and Remote Sensing Symposium, IGARSS 2018, Valencia, Spain, 22/07/18 - 27/07/18, pp. 3579-3582. https://doi.org/10.1109/IGARSS.2018.8518227

Global spatial and local spectral similarity-based group sparse representation for hyperspectral imagery classification. / Yu, Haoyang; Gao, Lianru; Liao, Wenzhi; Gamba, Paolo; Zhang, Bing.

2018. 3579-3582 Paper presented at 38th Annual IEEE International Geoscience and Remote Sensing Symposium, IGARSS 2018, Valencia, Spain.

Research output: Contribution to conferencePaper

TY - CONF

T1 - Global spatial and local spectral similarity-based group sparse representation for hyperspectral imagery classification

AU - Yu, Haoyang

AU - Gao, Lianru

AU - Liao, Wenzhi

AU - Gamba, Paolo

AU - Zhang, Bing

PY - 2018/11/5

Y1 - 2018/11/5

N2 - Spectral-spatial classification has been widely exploited for hyperspectral imagery. However, current methods either focus on local spatial similarity or global nonlocal self-similarity (NLSS). In this paper, we propose novel methods to couple both global spatial similarity and local spectral similarity together in a single framework. In particular, our approaches exploit global spatial similarity by searching non-overlap nonlocal patches, whereas spectral similarity is determined locally within the found patches. Experimental results on two real hyperspectral data sets demonstrate the efficiency of the proposed methods, with 5%-7% (overall classification accuracy) improvements over approaches that only consider either global or local similarity.

AB - Spectral-spatial classification has been widely exploited for hyperspectral imagery. However, current methods either focus on local spatial similarity or global nonlocal self-similarity (NLSS). In this paper, we propose novel methods to couple both global spatial similarity and local spectral similarity together in a single framework. In particular, our approaches exploit global spatial similarity by searching non-overlap nonlocal patches, whereas spectral similarity is determined locally within the found patches. Experimental results on two real hyperspectral data sets demonstrate the efficiency of the proposed methods, with 5%-7% (overall classification accuracy) improvements over approaches that only consider either global or local similarity.

KW - hyperspectral image

KW - classification

KW - nonlocal self-similarity

KW - group sparse representation

KW - support vector machines

KW - spatial coherence

KW - geophysical image processing

UR - http://hdl.handle.net/1854/LU-8565502

U2 - 10.1109/IGARSS.2018.8518227

DO - 10.1109/IGARSS.2018.8518227

M3 - Paper

SP - 3579

EP - 3582

ER -

Yu H, Gao L, Liao W, Gamba P, Zhang B. Global spatial and local spectral similarity-based group sparse representation for hyperspectral imagery classification. 2018. Paper presented at 38th Annual IEEE International Geoscience and Remote Sensing Symposium, IGARSS 2018, Valencia, Spain. https://doi.org/10.1109/IGARSS.2018.8518227