Segmented autoencoders for unsupervised embedded hyperspectral band selection

Research output: Contribution to conferencePaper

20 Downloads (Pure)

Abstract

One of the major challenges in hyperspectral imaging (HSI) is the selection of the most informative wavelengths within the vast amount of data in a hypercube. Band selection can reduce the amount of data and computational cost as well as counteracting the negative effects of redundant and erroneous information. In this paper, we propose an unsupervised, embedded band selection algorithm that utilises the deep learning framework. Autoencoders are used to reconstruct measured spectral signatures. By putting a sparsity constraint on the input weights, the bands that contribute most to the reconstruction can be identified and chosen as the selected bands. Additionally, segmenting the input data into several spectral regions and distributing the number of desired bands according to a density measure among these segments, the quality of the selected bands can be increased and the computational time reduced by training several autoencoders. Results on a benchmark remote sensing HSI dataset show that the proposed algorithm improves classification accuracy compared to other state of the art band selection algorithms and thereby builds the basis for a framework of embedded band selection in HSI.
Original languageEnglish
Number of pages6
Publication statusPublished - 26 Nov 2018
Event7th European Workshop on Visual Information Processing - Tampere, Finland
Duration: 26 Nov 201828 Nov 2018

Conference

Conference7th European Workshop on Visual Information Processing
Abbreviated titleEUVIP
CountryFinland
CityTampere
Period26/11/1828/11/18

Fingerprint

Remote sensing
Wavelength
Hyperspectral imaging
Costs
Deep learning

Keywords

  • hyperspectral imaging
  • autoencoder
  • band selection

Cite this

Tschannerl, J., Ren, J., Zabalza, J., & Marshall, S. (2018). Segmented autoencoders for unsupervised embedded hyperspectral band selection. Paper presented at 7th European Workshop on Visual Information Processing, Tampere, Finland.
Tschannerl, Julius ; Ren, Jinchang ; Zabalza, Jaime ; Marshall, Stephen. / Segmented autoencoders for unsupervised embedded hyperspectral band selection. Paper presented at 7th European Workshop on Visual Information Processing, Tampere, Finland.6 p.
@conference{4fe76932d7c640ab9b70cbec6aead58f,
title = "Segmented autoencoders for unsupervised embedded hyperspectral band selection",
abstract = "One of the major challenges in hyperspectral imaging (HSI) is the selection of the most informative wavelengths within the vast amount of data in a hypercube. Band selection can reduce the amount of data and computational cost as well as counteracting the negative effects of redundant and erroneous information. In this paper, we propose an unsupervised, embedded band selection algorithm that utilises the deep learning framework. Autoencoders are used to reconstruct measured spectral signatures. By putting a sparsity constraint on the input weights, the bands that contribute most to the reconstruction can be identified and chosen as the selected bands. Additionally, segmenting the input data into several spectral regions and distributing the number of desired bands according to a density measure among these segments, the quality of the selected bands can be increased and the computational time reduced by training several autoencoders. Results on a benchmark remote sensing HSI dataset show that the proposed algorithm improves classification accuracy compared to other state of the art band selection algorithms and thereby builds the basis for a framework of embedded band selection in HSI.",
keywords = "hyperspectral imaging, autoencoder, band selection",
author = "Julius Tschannerl and Jinchang Ren and Jaime Zabalza and Stephen Marshall",
note = "{\circledC} 2018 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works.; 7th European Workshop on Visual Information Processing, EUVIP ; Conference date: 26-11-2018 Through 28-11-2018",
year = "2018",
month = "11",
day = "26",
language = "English",

}

Tschannerl, J, Ren, J, Zabalza, J & Marshall, S 2018, 'Segmented autoencoders for unsupervised embedded hyperspectral band selection' Paper presented at 7th European Workshop on Visual Information Processing, Tampere, Finland, 26/11/18 - 28/11/18, .

Segmented autoencoders for unsupervised embedded hyperspectral band selection. / Tschannerl, Julius; Ren, Jinchang; Zabalza, Jaime; Marshall, Stephen.

2018. Paper presented at 7th European Workshop on Visual Information Processing, Tampere, Finland.

Research output: Contribution to conferencePaper

TY - CONF

T1 - Segmented autoencoders for unsupervised embedded hyperspectral band selection

AU - Tschannerl, Julius

AU - Ren, Jinchang

AU - Zabalza, Jaime

AU - Marshall, Stephen

N1 - © 2018 IEEE. Personal use of this material is permitted. Permission from IEEE must be obtained for all other uses, in any current or future media, including reprinting/republishing this material for advertising or promotional purposes, creating new collective works, for resale or redistribution to servers or lists, or reuse of any copyrighted component of this work in other works.

PY - 2018/11/26

Y1 - 2018/11/26

N2 - One of the major challenges in hyperspectral imaging (HSI) is the selection of the most informative wavelengths within the vast amount of data in a hypercube. Band selection can reduce the amount of data and computational cost as well as counteracting the negative effects of redundant and erroneous information. In this paper, we propose an unsupervised, embedded band selection algorithm that utilises the deep learning framework. Autoencoders are used to reconstruct measured spectral signatures. By putting a sparsity constraint on the input weights, the bands that contribute most to the reconstruction can be identified and chosen as the selected bands. Additionally, segmenting the input data into several spectral regions and distributing the number of desired bands according to a density measure among these segments, the quality of the selected bands can be increased and the computational time reduced by training several autoencoders. Results on a benchmark remote sensing HSI dataset show that the proposed algorithm improves classification accuracy compared to other state of the art band selection algorithms and thereby builds the basis for a framework of embedded band selection in HSI.

AB - One of the major challenges in hyperspectral imaging (HSI) is the selection of the most informative wavelengths within the vast amount of data in a hypercube. Band selection can reduce the amount of data and computational cost as well as counteracting the negative effects of redundant and erroneous information. In this paper, we propose an unsupervised, embedded band selection algorithm that utilises the deep learning framework. Autoencoders are used to reconstruct measured spectral signatures. By putting a sparsity constraint on the input weights, the bands that contribute most to the reconstruction can be identified and chosen as the selected bands. Additionally, segmenting the input data into several spectral regions and distributing the number of desired bands according to a density measure among these segments, the quality of the selected bands can be increased and the computational time reduced by training several autoencoders. Results on a benchmark remote sensing HSI dataset show that the proposed algorithm improves classification accuracy compared to other state of the art band selection algorithms and thereby builds the basis for a framework of embedded band selection in HSI.

KW - hyperspectral imaging

KW - autoencoder

KW - band selection

UR - http://www.tut.fi/euvip2018/index.html

M3 - Paper

ER -

Tschannerl J, Ren J, Zabalza J, Marshall S. Segmented autoencoders for unsupervised embedded hyperspectral band selection. 2018. Paper presented at 7th European Workshop on Visual Information Processing, Tampere, Finland.