No Cover Image

Journal article 1183 views 254 downloads

FSPE: Visualization of Hyperspectral Imagery Using Faithful Stochastic Proximity Embedding

Safa A. Najim, Ik Soo Lim, Peter Wittek, Mark Jones Orcid Logo

IEEE Geoscience and Remote Sensing Letters, Volume: 12, Issue: 1, Pages: 18 - 22

Swansea University Author: Mark Jones Orcid Logo

DOI (Published version): 10.1109/LGRS.2014.2324631

Abstract

Hyperspectral image visualization reduces color bands to three, but prevailing linear methods fail to address data characteristics, and nonlinear embeddings are computationally demanding. Qualitative evaluation of embedding is also lacking. We propose faithful stochastic proximity embedding (FSPE),...

Full description

Published in: IEEE Geoscience and Remote Sensing Letters
Published: 2014
URI: https://cronfa.swan.ac.uk/Record/cronfa18053
Tags: Add Tag
No Tags, Be the first to tag this record!
first_indexed 2014-06-24T01:59:49Z
last_indexed 2019-06-21T13:28:10Z
id cronfa18053
recordtype SURis
fullrecord <?xml version="1.0"?><rfc1807><datestamp>2019-06-21T12:45:35.1032723</datestamp><bib-version>v2</bib-version><id>18053</id><entry>2014-06-23</entry><title>FSPE: Visualization of Hyperspectral Imagery Using Faithful Stochastic Proximity Embedding</title><swanseaauthors><author><sid>2e1030b6e14fc9debd5d5ae7cc335562</sid><ORCID>0000-0001-8991-1190</ORCID><firstname>Mark</firstname><surname>Jones</surname><name>Mark Jones</name><active>true</active><ethesisStudent>false</ethesisStudent></author></swanseaauthors><date>2014-06-23</date><deptcode>SCS</deptcode><abstract>Hyperspectral image visualization reduces color bands to three, but prevailing linear methods fail to address data characteristics, and nonlinear embeddings are computationally demanding. Qualitative evaluation of embedding is also lacking. We propose faithful stochastic proximity embedding (FSPE), which is a scalable and nonlinear dimensionality reduction method. FSPE considers the nonlinear characteristics of spectral signatures, yet it avoids the costly computation of geodesic distances that are often required by other nonlinear methods. Furthermore, we employ a pixelwise metric that measures the quality of hyperspectral image visualization at each pixel. FSPE outperforms the state-of-art methods by at least 12% on average and up to 25% in the qualitative measure. An implementation on graphics processing units is two orders of magnitude faster than the baseline. Our method opens the path to high-fidelity and real-time analysis of hyperspectral images.</abstract><type>Journal Article</type><journal>IEEE Geoscience and Remote Sensing Letters</journal><volume>12</volume><journalNumber>1</journalNumber><paginationStart>18</paginationStart><paginationEnd>22</paginationEnd><publisher/><keywords/><publishedDay>20</publishedDay><publishedMonth>6</publishedMonth><publishedYear>2014</publishedYear><publishedDate>2014-06-20</publishedDate><doi>10.1109/LGRS.2014.2324631</doi><url/><notes></notes><college>COLLEGE NANME</college><department>Computer Science</department><CollegeCode>COLLEGE CODE</CollegeCode><DepartmentCode>SCS</DepartmentCode><institution>Swansea University</institution><apcterm/><lastEdited>2019-06-21T12:45:35.1032723</lastEdited><Created>2014-06-23T12:43:41.3024372</Created><path><level id="1">Faculty of Science and Engineering</level><level id="2">School of Mathematics and Computer Science - Computer Science</level></path><authors><author><firstname>Safa A.</firstname><surname>Najim</surname><order>1</order></author><author><firstname>Ik Soo</firstname><surname>Lim</surname><order>2</order></author><author><firstname>Peter</firstname><surname>Wittek</surname><order>3</order></author><author><firstname>Mark</firstname><surname>Jones</surname><orcid>0000-0001-8991-1190</orcid><order>4</order></author></authors><documents><document><filename>0018053-15042015121236.pdf</filename><originalFilename>FSPE.pdf</originalFilename><uploaded>2015-04-15T12:12:36.1570000</uploaded><type>Output</type><contentLength>1168369</contentLength><contentType>application/pdf</contentType><version>Version of Record</version><cronfaStatus>true</cronfaStatus><embargoDate>2015-04-14T00:00:00.0000000</embargoDate><documentNotes/><copyrightCorrect>true</copyrightCorrect></document></documents><OutputDurs/></rfc1807>
spelling 2019-06-21T12:45:35.1032723 v2 18053 2014-06-23 FSPE: Visualization of Hyperspectral Imagery Using Faithful Stochastic Proximity Embedding 2e1030b6e14fc9debd5d5ae7cc335562 0000-0001-8991-1190 Mark Jones Mark Jones true false 2014-06-23 SCS Hyperspectral image visualization reduces color bands to three, but prevailing linear methods fail to address data characteristics, and nonlinear embeddings are computationally demanding. Qualitative evaluation of embedding is also lacking. We propose faithful stochastic proximity embedding (FSPE), which is a scalable and nonlinear dimensionality reduction method. FSPE considers the nonlinear characteristics of spectral signatures, yet it avoids the costly computation of geodesic distances that are often required by other nonlinear methods. Furthermore, we employ a pixelwise metric that measures the quality of hyperspectral image visualization at each pixel. FSPE outperforms the state-of-art methods by at least 12% on average and up to 25% in the qualitative measure. An implementation on graphics processing units is two orders of magnitude faster than the baseline. Our method opens the path to high-fidelity and real-time analysis of hyperspectral images. Journal Article IEEE Geoscience and Remote Sensing Letters 12 1 18 22 20 6 2014 2014-06-20 10.1109/LGRS.2014.2324631 COLLEGE NANME Computer Science COLLEGE CODE SCS Swansea University 2019-06-21T12:45:35.1032723 2014-06-23T12:43:41.3024372 Faculty of Science and Engineering School of Mathematics and Computer Science - Computer Science Safa A. Najim 1 Ik Soo Lim 2 Peter Wittek 3 Mark Jones 0000-0001-8991-1190 4 0018053-15042015121236.pdf FSPE.pdf 2015-04-15T12:12:36.1570000 Output 1168369 application/pdf Version of Record true 2015-04-14T00:00:00.0000000 true
title FSPE: Visualization of Hyperspectral Imagery Using Faithful Stochastic Proximity Embedding
spellingShingle FSPE: Visualization of Hyperspectral Imagery Using Faithful Stochastic Proximity Embedding
Mark Jones
title_short FSPE: Visualization of Hyperspectral Imagery Using Faithful Stochastic Proximity Embedding
title_full FSPE: Visualization of Hyperspectral Imagery Using Faithful Stochastic Proximity Embedding
title_fullStr FSPE: Visualization of Hyperspectral Imagery Using Faithful Stochastic Proximity Embedding
title_full_unstemmed FSPE: Visualization of Hyperspectral Imagery Using Faithful Stochastic Proximity Embedding
title_sort FSPE: Visualization of Hyperspectral Imagery Using Faithful Stochastic Proximity Embedding
author_id_str_mv 2e1030b6e14fc9debd5d5ae7cc335562
author_id_fullname_str_mv 2e1030b6e14fc9debd5d5ae7cc335562_***_Mark Jones
author Mark Jones
author2 Safa A. Najim
Ik Soo Lim
Peter Wittek
Mark Jones
format Journal article
container_title IEEE Geoscience and Remote Sensing Letters
container_volume 12
container_issue 1
container_start_page 18
publishDate 2014
institution Swansea University
doi_str_mv 10.1109/LGRS.2014.2324631
college_str Faculty of Science and Engineering
hierarchytype
hierarchy_top_id facultyofscienceandengineering
hierarchy_top_title Faculty of Science and Engineering
hierarchy_parent_id facultyofscienceandengineering
hierarchy_parent_title Faculty of Science and Engineering
department_str School of Mathematics and Computer Science - Computer Science{{{_:::_}}}Faculty of Science and Engineering{{{_:::_}}}School of Mathematics and Computer Science - Computer Science
document_store_str 1
active_str 0
description Hyperspectral image visualization reduces color bands to three, but prevailing linear methods fail to address data characteristics, and nonlinear embeddings are computationally demanding. Qualitative evaluation of embedding is also lacking. We propose faithful stochastic proximity embedding (FSPE), which is a scalable and nonlinear dimensionality reduction method. FSPE considers the nonlinear characteristics of spectral signatures, yet it avoids the costly computation of geodesic distances that are often required by other nonlinear methods. Furthermore, we employ a pixelwise metric that measures the quality of hyperspectral image visualization at each pixel. FSPE outperforms the state-of-art methods by at least 12% on average and up to 25% in the qualitative measure. An implementation on graphics processing units is two orders of magnitude faster than the baseline. Our method opens the path to high-fidelity and real-time analysis of hyperspectral images.
published_date 2014-06-20T03:21:03Z
_version_ 1763750616962695168
score 10.988012