Superpixel based feature specific sparse representation for spectral-spatial classification of hyperspectral images

Research output: Contribution to journalArticle

2 Citations (Scopus)
12 Downloads (Pure)

Abstract

To improve the performance of the sparse representation classification (SRC), we propose a superpixel-based feature specific sparse representation framework (SPFS-SRC) for spectral-spatial classification of hyperspectral images (HSI) at superpixel level. First, the HSI is divided into different spatial regions, each region is shape- and size-adapted and considered as a superpixel. For each superpixel, it contains a number of pixels with similar spectral characteristic. Since the utilization of multiple features in HSI classification has been proved to be an effective strategy, we have generated both spatial and spectral features for each superpixel. By assuming that all the pixels in a superpixel belongs to one certain class, a kernel SRC is introduced to the classification of HSI. In the SRC framework, we have employed a metric learning strategy to exploit the commonalities of different features. Experimental results on two popular HSI datasets have demonstrated the efficacy of our proposed methodology.

Original languageEnglish
Article number536
Number of pages17
JournalRemote Sensing
Volume11
Issue number5
DOIs
Publication statusPublished - 5 Mar 2019

Fingerprint

pixel
Pixels
Image classification
image classification
learning
methodology

Keywords

  • hyperspectral image
  • image classification
  • metric learning
  • sparse representation
  • superpixel

Cite this

@article{771eb011409e4fecb68e8f8361cb1cdb,
title = "Superpixel based feature specific sparse representation for spectral-spatial classification of hyperspectral images",
abstract = "To improve the performance of the sparse representation classification (SRC), we propose a superpixel-based feature specific sparse representation framework (SPFS-SRC) for spectral-spatial classification of hyperspectral images (HSI) at superpixel level. First, the HSI is divided into different spatial regions, each region is shape- and size-adapted and considered as a superpixel. For each superpixel, it contains a number of pixels with similar spectral characteristic. Since the utilization of multiple features in HSI classification has been proved to be an effective strategy, we have generated both spatial and spectral features for each superpixel. By assuming that all the pixels in a superpixel belongs to one certain class, a kernel SRC is introduced to the classification of HSI. In the SRC framework, we have employed a metric learning strategy to exploit the commonalities of different features. Experimental results on two popular HSI datasets have demonstrated the efficacy of our proposed methodology.",
keywords = "hyperspectral image, image classification, metric learning, sparse representation, superpixel",
author = "He Sun and Jinchang Ren and Huimin Zhao and Yijun Yan and Jaime Zabalza and Stephen Marshall",
year = "2019",
month = "3",
day = "5",
doi = "10.3390/rs11050536",
language = "English",
volume = "11",
journal = "Remote Sensing",
issn = "2072-4292",
number = "5",

}

TY - JOUR

T1 - Superpixel based feature specific sparse representation for spectral-spatial classification of hyperspectral images

AU - Sun, He

AU - Ren, Jinchang

AU - Zhao, Huimin

AU - Yan, Yijun

AU - Zabalza, Jaime

AU - Marshall, Stephen

PY - 2019/3/5

Y1 - 2019/3/5

N2 - To improve the performance of the sparse representation classification (SRC), we propose a superpixel-based feature specific sparse representation framework (SPFS-SRC) for spectral-spatial classification of hyperspectral images (HSI) at superpixel level. First, the HSI is divided into different spatial regions, each region is shape- and size-adapted and considered as a superpixel. For each superpixel, it contains a number of pixels with similar spectral characteristic. Since the utilization of multiple features in HSI classification has been proved to be an effective strategy, we have generated both spatial and spectral features for each superpixel. By assuming that all the pixels in a superpixel belongs to one certain class, a kernel SRC is introduced to the classification of HSI. In the SRC framework, we have employed a metric learning strategy to exploit the commonalities of different features. Experimental results on two popular HSI datasets have demonstrated the efficacy of our proposed methodology.

AB - To improve the performance of the sparse representation classification (SRC), we propose a superpixel-based feature specific sparse representation framework (SPFS-SRC) for spectral-spatial classification of hyperspectral images (HSI) at superpixel level. First, the HSI is divided into different spatial regions, each region is shape- and size-adapted and considered as a superpixel. For each superpixel, it contains a number of pixels with similar spectral characteristic. Since the utilization of multiple features in HSI classification has been proved to be an effective strategy, we have generated both spatial and spectral features for each superpixel. By assuming that all the pixels in a superpixel belongs to one certain class, a kernel SRC is introduced to the classification of HSI. In the SRC framework, we have employed a metric learning strategy to exploit the commonalities of different features. Experimental results on two popular HSI datasets have demonstrated the efficacy of our proposed methodology.

KW - hyperspectral image

KW - image classification

KW - metric learning

KW - sparse representation

KW - superpixel

UR - http://www.scopus.com/inward/record.url?scp=85063011784&partnerID=8YFLogxK

UR - https://www.mdpi.com/journal/remotesensing

U2 - 10.3390/rs11050536

DO - 10.3390/rs11050536

M3 - Article

VL - 11

JO - Remote Sensing

JF - Remote Sensing

SN - 2072-4292

IS - 5

M1 - 536

ER -