Dimensionality reduction based on determinantal point process and singular spectrum analysis for hyperspectral images

Weizhao Chen, Zhijing Yang, Faxian Cao, Yijun Yan, Meilin Wang, Chunmei Qing, Yongqiang Cheng

Research output: Contribution to journalArticle

2 Citations (Scopus)

Abstract

Dimensionality reduction is of high importance in hyperspectral data processing, which can effectively reduce the data redundancy and computation time for improved classification accuracy. Band selection and feature extraction methods are two widely used dimensionality reduction techniques. By integrating the advantages of the band selection and feature extraction, the authors propose a new method for reducing the dimension of hyperspectral image data. First, a new and fast band selection algorithm is proposed for hyperspectral images based on an improved determinantal point process (DPP). To reduce the amount of calculation, the dual-DPP is used for fast sampling representative pixels, followed by k-nearest neighbour-based local processing to explore more spatial information. These representative pixel points are used to construct multiple adjacency matrices to describe the correlation between bands based on mutual information. To further improve the classification accuracy, two-dimensional singular spectrum analysis is used for feature extraction from the selected bands. Experiments show that the proposed method can select a low-redundancy and representative band subset, where both data dimension and computation time can be reduced. Furthermore, it also shows that the proposed dimensionality reduction algorithm outperforms a number of state-of-the-art methods in terms of classification accuracy.
LanguageEnglish
JournalIET Image Processing
Early online date5 Sep 2018
DOIs
Publication statusE-pub ahead of print - 5 Sep 2018

Fingerprint

Spectrum analysis
Feature extraction
Redundancy
Pixels
Set theory
Sampling
Processing
Experiments

Keywords

  • image representation
  • spectral analysis
  • feature extraction
  • image sampling
  • geophysical image processing
  • hyperspectral imaging
  • image classification
  • learning

Cite this

Chen, Weizhao ; Yang, Zhijing ; Cao, Faxian ; Yan, Yijun ; Wang, Meilin ; Qing, Chunmei ; Cheng, Yongqiang. / Dimensionality reduction based on determinantal point process and singular spectrum analysis for hyperspectral images. In: IET Image Processing. 2018.
@article{7bea0fca865344818e26fd55c29aefcf,
title = "Dimensionality reduction based on determinantal point process and singular spectrum analysis for hyperspectral images",
abstract = "Dimensionality reduction is of high importance in hyperspectral data processing, which can effectively reduce the data redundancy and computation time for improved classification accuracy. Band selection and feature extraction methods are two widely used dimensionality reduction techniques. By integrating the advantages of the band selection and feature extraction, the authors propose a new method for reducing the dimension of hyperspectral image data. First, a new and fast band selection algorithm is proposed for hyperspectral images based on an improved determinantal point process (DPP). To reduce the amount of calculation, the dual-DPP is used for fast sampling representative pixels, followed by k-nearest neighbour-based local processing to explore more spatial information. These representative pixel points are used to construct multiple adjacency matrices to describe the correlation between bands based on mutual information. To further improve the classification accuracy, two-dimensional singular spectrum analysis is used for feature extraction from the selected bands. Experiments show that the proposed method can select a low-redundancy and representative band subset, where both data dimension and computation time can be reduced. Furthermore, it also shows that the proposed dimensionality reduction algorithm outperforms a number of state-of-the-art methods in terms of classification accuracy.",
keywords = "image representation, spectral analysis, feature extraction, image sampling, geophysical image processing, hyperspectral imaging, image classification, learning",
author = "Weizhao Chen and Zhijing Yang and Faxian Cao and Yijun Yan and Meilin Wang and Chunmei Qing and Yongqiang Cheng",
year = "2018",
month = "9",
day = "5",
doi = "10.1049/iet-ipr.2018.5419",
language = "English",
journal = "IET Image Processing",
issn = "1751-9659",

}

Dimensionality reduction based on determinantal point process and singular spectrum analysis for hyperspectral images. / Chen, Weizhao; Yang, Zhijing; Cao, Faxian; Yan, Yijun; Wang, Meilin; Qing, Chunmei; Cheng, Yongqiang.

In: IET Image Processing, 05.09.2018.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Dimensionality reduction based on determinantal point process and singular spectrum analysis for hyperspectral images

AU - Chen, Weizhao

AU - Yang, Zhijing

AU - Cao, Faxian

AU - Yan, Yijun

AU - Wang, Meilin

AU - Qing, Chunmei

AU - Cheng, Yongqiang

PY - 2018/9/5

Y1 - 2018/9/5

N2 - Dimensionality reduction is of high importance in hyperspectral data processing, which can effectively reduce the data redundancy and computation time for improved classification accuracy. Band selection and feature extraction methods are two widely used dimensionality reduction techniques. By integrating the advantages of the band selection and feature extraction, the authors propose a new method for reducing the dimension of hyperspectral image data. First, a new and fast band selection algorithm is proposed for hyperspectral images based on an improved determinantal point process (DPP). To reduce the amount of calculation, the dual-DPP is used for fast sampling representative pixels, followed by k-nearest neighbour-based local processing to explore more spatial information. These representative pixel points are used to construct multiple adjacency matrices to describe the correlation between bands based on mutual information. To further improve the classification accuracy, two-dimensional singular spectrum analysis is used for feature extraction from the selected bands. Experiments show that the proposed method can select a low-redundancy and representative band subset, where both data dimension and computation time can be reduced. Furthermore, it also shows that the proposed dimensionality reduction algorithm outperforms a number of state-of-the-art methods in terms of classification accuracy.

AB - Dimensionality reduction is of high importance in hyperspectral data processing, which can effectively reduce the data redundancy and computation time for improved classification accuracy. Band selection and feature extraction methods are two widely used dimensionality reduction techniques. By integrating the advantages of the band selection and feature extraction, the authors propose a new method for reducing the dimension of hyperspectral image data. First, a new and fast band selection algorithm is proposed for hyperspectral images based on an improved determinantal point process (DPP). To reduce the amount of calculation, the dual-DPP is used for fast sampling representative pixels, followed by k-nearest neighbour-based local processing to explore more spatial information. These representative pixel points are used to construct multiple adjacency matrices to describe the correlation between bands based on mutual information. To further improve the classification accuracy, two-dimensional singular spectrum analysis is used for feature extraction from the selected bands. Experiments show that the proposed method can select a low-redundancy and representative band subset, where both data dimension and computation time can be reduced. Furthermore, it also shows that the proposed dimensionality reduction algorithm outperforms a number of state-of-the-art methods in terms of classification accuracy.

KW - image representation

KW - spectral analysis

KW - feature extraction

KW - image sampling

KW - geophysical image processing

KW - hyperspectral imaging

KW - image classification

KW - learning

UR - http://digital-library.theiet.org/content/journals/iet-ipr

U2 - 10.1049/iet-ipr.2018.5419

DO - 10.1049/iet-ipr.2018.5419

M3 - Article

JO - IET Image Processing

T2 - IET Image Processing

JF - IET Image Processing

SN - 1751-9659

ER -