Spectral matching accuracy in processing hyperspectral data

Stefan A. Robila, Andrew Gershman

Research output: Chapter in Book/Report/Conference proceedingConference contribution

45 Citations (Scopus)

Abstract

We investigated the accuracy of several spectral metrics when used for matching in hyperspectral imagery. Spectral matching refers to measuring the similarity among spectra. Two spectra are similar when the spectra distance between them is small and dissimilar otherwise. Spectral matching is used in many areas such as target detection and spectra classification. At the heart of spectral matching lay the distance measure used and the threshold value differentiating between similar and dissimilar. Frequent choices for such measures are Spectral Angle (SA) and Spectral Information Divergence (SID) since they provide a limited range for threshold values. In addition, two new measures Spectral Correlation Angle (SGA) and Spectral Gradient Angle (SGA) are developed. Next, we suggest an alternative measure developed by using a Normalized Euclidean Distance (NED). To lest the accuracy of the measures we used target information extracted from HYDICE data, and assessment tools such as the relative spectral discriminatory probability and the relative spectral discriminatory entropy. Experimental results suggest that NED outperforms SA, SCA and SGA and is relatively equivalent to SID. Given the reduction In computation time, NED constitutes an attractive measure to be used in spectra matching.

Original languageEnglish
Title of host publicationISSCS 2005
Subtitle of host publicationInternational Symposium on Signals, Circuits and Systems - Proceedings
Pages163-166
Number of pages4
DOIs
StatePublished - 1 Dec 2005
EventISSCS 2005: International Symposium on Signals, Circuits and Systems - Iasi, Romania
Duration: 14 Jul 200515 Jul 2005

Publication series

NameISSCS 2005: International Symposium on Signals, Circuits and Systems - Proceedings
Volume1

Other

OtherISSCS 2005: International Symposium on Signals, Circuits and Systems
CountryRomania
CityIasi
Period14/07/0515/07/05

Fingerprint

Target tracking
Entropy
Processing

Keywords

  • Euclidean distance
  • Hyperspectral images
  • Spectral angle
  • Spectral gradient angle
  • Spectral information divergence

Cite this

Robila, S. A., & Gershman, A. (2005). Spectral matching accuracy in processing hyperspectral data. In ISSCS 2005: International Symposium on Signals, Circuits and Systems - Proceedings (pp. 163-166). [1509878] (ISSCS 2005: International Symposium on Signals, Circuits and Systems - Proceedings; Vol. 1). https://doi.org/10.1109/ISSCS.2005.1509878
Robila, Stefan A. ; Gershman, Andrew. / Spectral matching accuracy in processing hyperspectral data. ISSCS 2005: International Symposium on Signals, Circuits and Systems - Proceedings. 2005. pp. 163-166 (ISSCS 2005: International Symposium on Signals, Circuits and Systems - Proceedings).
@inproceedings{acf0ef16d0614dd48bd4a6d6ea5a44c7,
title = "Spectral matching accuracy in processing hyperspectral data",
abstract = "We investigated the accuracy of several spectral metrics when used for matching in hyperspectral imagery. Spectral matching refers to measuring the similarity among spectra. Two spectra are similar when the spectra distance between them is small and dissimilar otherwise. Spectral matching is used in many areas such as target detection and spectra classification. At the heart of spectral matching lay the distance measure used and the threshold value differentiating between similar and dissimilar. Frequent choices for such measures are Spectral Angle (SA) and Spectral Information Divergence (SID) since they provide a limited range for threshold values. In addition, two new measures Spectral Correlation Angle (SGA) and Spectral Gradient Angle (SGA) are developed. Next, we suggest an alternative measure developed by using a Normalized Euclidean Distance (NED). To lest the accuracy of the measures we used target information extracted from HYDICE data, and assessment tools such as the relative spectral discriminatory probability and the relative spectral discriminatory entropy. Experimental results suggest that NED outperforms SA, SCA and SGA and is relatively equivalent to SID. Given the reduction In computation time, NED constitutes an attractive measure to be used in spectra matching.",
keywords = "Euclidean distance, Hyperspectral images, Spectral angle, Spectral gradient angle, Spectral information divergence",
author = "Robila, {Stefan A.} and Andrew Gershman",
year = "2005",
month = "12",
day = "1",
doi = "10.1109/ISSCS.2005.1509878",
language = "English",
isbn = "0780390296",
series = "ISSCS 2005: International Symposium on Signals, Circuits and Systems - Proceedings",
pages = "163--166",
booktitle = "ISSCS 2005",

}

Robila, SA & Gershman, A 2005, Spectral matching accuracy in processing hyperspectral data. in ISSCS 2005: International Symposium on Signals, Circuits and Systems - Proceedings., 1509878, ISSCS 2005: International Symposium on Signals, Circuits and Systems - Proceedings, vol. 1, pp. 163-166, ISSCS 2005: International Symposium on Signals, Circuits and Systems, Iasi, Romania, 14/07/05. https://doi.org/10.1109/ISSCS.2005.1509878

Spectral matching accuracy in processing hyperspectral data. / Robila, Stefan A.; Gershman, Andrew.

ISSCS 2005: International Symposium on Signals, Circuits and Systems - Proceedings. 2005. p. 163-166 1509878 (ISSCS 2005: International Symposium on Signals, Circuits and Systems - Proceedings; Vol. 1).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

TY - GEN

T1 - Spectral matching accuracy in processing hyperspectral data

AU - Robila, Stefan A.

AU - Gershman, Andrew

PY - 2005/12/1

Y1 - 2005/12/1

N2 - We investigated the accuracy of several spectral metrics when used for matching in hyperspectral imagery. Spectral matching refers to measuring the similarity among spectra. Two spectra are similar when the spectra distance between them is small and dissimilar otherwise. Spectral matching is used in many areas such as target detection and spectra classification. At the heart of spectral matching lay the distance measure used and the threshold value differentiating between similar and dissimilar. Frequent choices for such measures are Spectral Angle (SA) and Spectral Information Divergence (SID) since they provide a limited range for threshold values. In addition, two new measures Spectral Correlation Angle (SGA) and Spectral Gradient Angle (SGA) are developed. Next, we suggest an alternative measure developed by using a Normalized Euclidean Distance (NED). To lest the accuracy of the measures we used target information extracted from HYDICE data, and assessment tools such as the relative spectral discriminatory probability and the relative spectral discriminatory entropy. Experimental results suggest that NED outperforms SA, SCA and SGA and is relatively equivalent to SID. Given the reduction In computation time, NED constitutes an attractive measure to be used in spectra matching.

AB - We investigated the accuracy of several spectral metrics when used for matching in hyperspectral imagery. Spectral matching refers to measuring the similarity among spectra. Two spectra are similar when the spectra distance between them is small and dissimilar otherwise. Spectral matching is used in many areas such as target detection and spectra classification. At the heart of spectral matching lay the distance measure used and the threshold value differentiating between similar and dissimilar. Frequent choices for such measures are Spectral Angle (SA) and Spectral Information Divergence (SID) since they provide a limited range for threshold values. In addition, two new measures Spectral Correlation Angle (SGA) and Spectral Gradient Angle (SGA) are developed. Next, we suggest an alternative measure developed by using a Normalized Euclidean Distance (NED). To lest the accuracy of the measures we used target information extracted from HYDICE data, and assessment tools such as the relative spectral discriminatory probability and the relative spectral discriminatory entropy. Experimental results suggest that NED outperforms SA, SCA and SGA and is relatively equivalent to SID. Given the reduction In computation time, NED constitutes an attractive measure to be used in spectra matching.

KW - Euclidean distance

KW - Hyperspectral images

KW - Spectral angle

KW - Spectral gradient angle

KW - Spectral information divergence

UR - http://www.scopus.com/inward/record.url?scp=33749070553&partnerID=8YFLogxK

U2 - 10.1109/ISSCS.2005.1509878

DO - 10.1109/ISSCS.2005.1509878

M3 - Conference contribution

AN - SCOPUS:33749070553

SN - 0780390296

SN - 9780780390294

T3 - ISSCS 2005: International Symposium on Signals, Circuits and Systems - Proceedings

SP - 163

EP - 166

BT - ISSCS 2005

ER -

Robila SA, Gershman A. Spectral matching accuracy in processing hyperspectral data. In ISSCS 2005: International Symposium on Signals, Circuits and Systems - Proceedings. 2005. p. 163-166. 1509878. (ISSCS 2005: International Symposium on Signals, Circuits and Systems - Proceedings). https://doi.org/10.1109/ISSCS.2005.1509878