A decision-level fusion scheme using the support vector data description for target detection in hyperspectral imagery

Wesam A. Sakla, Adel A. Sakla, Andrew Chan, Jim Ji

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

Abstract

Spectral variability remains a challenging problem for target detection in hyperspectral (HS) imagery. In previous work, we developed a target detection scheme using the kernel-based support vector data description (SVDD). We constructed a first-order Markov-based Gaussian model to generate samples to describe the spectral variability of the target class. However, the Gaussian-generated samples also require selection of the variance parameter σ2 that dictates the level of variability in the generated target class signatures. In this work, we have investigated the use of decision-level fusion techniques for alleviating the problem of choosing a proper value of σ2. We have trained a collection of SVDDs with unique variance parameters σ2 for each of the target training sets and have investigated their combination using the traditional AND, OR, and majority vote (MV) decision-level rules. We have inserted target signatures into an urban HS scene with differing levels of spectral variability to explore the performance of the proposed scheme in these scenarios. Experiments show that the MV fusion rule is the best choice, providing relatively low false positive rates (FPR) while yielding high true positive rates (TPR). Detection results show that the proposed SVDD-based decision-level scheme using the MV fusion rule is highly accurate and yields higher true positive rates (TPR) and lower false positive rates (FPR) than the adaptive matched filter (AMF).

Original languageEnglish
Title of host publicationAutomatic Target Recognition XX; Acquisition, Tracking, Pointing, and Laser Systems Technologies XXIV; and Optical Pattern Recognition XXI
Volume7696
DOIs
Publication statusPublished - 2010
Externally publishedYes
EventAutomatic Target Recognition XX; Acquisition, Tracking, Pointing, and Laser Systems Technologies XXIV; and Optical Pattern Recognition XXI - Orlando, FL, United States
Duration: 5 Apr 20108 Apr 2010

Other

OtherAutomatic Target Recognition XX; Acquisition, Tracking, Pointing, and Laser Systems Technologies XXIV; and Optical Pattern Recognition XXI
CountryUnited States
CityOrlando, FL
Period5/4/108/4/10

Fingerprint

Support Vector Data Description
Data description
Hyperspectral Imagery
Target Detection
Target tracking
imagery
Fusion
Vote
Fusion reactions
fusion
Fusion Rule
Target
False Positive
Signature
Matched filters
Adaptive filters
Matched Filter
Adaptive Filter
Gaussian Model
signatures

Keywords

  • Automatic target recognition
  • Decision-level fusion
  • Hyperspectral imagery
  • Majority vote rule
  • Spectral variability
  • Support vector data description
  • Target detection

ASJC Scopus subject areas

  • Electronic, Optical and Magnetic Materials
  • Condensed Matter Physics
  • Computer Science Applications
  • Applied Mathematics
  • Electrical and Electronic Engineering

Cite this

Sakla, W. A., Sakla, A. A., Chan, A., & Ji, J. (2010). A decision-level fusion scheme using the support vector data description for target detection in hyperspectral imagery. In Automatic Target Recognition XX; Acquisition, Tracking, Pointing, and Laser Systems Technologies XXIV; and Optical Pattern Recognition XXI (Vol. 7696). [76960Y] https://doi.org/10.1117/12.849936

A decision-level fusion scheme using the support vector data description for target detection in hyperspectral imagery. / Sakla, Wesam A.; Sakla, Adel A.; Chan, Andrew; Ji, Jim.

Automatic Target Recognition XX; Acquisition, Tracking, Pointing, and Laser Systems Technologies XXIV; and Optical Pattern Recognition XXI. Vol. 7696 2010. 76960Y.

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

Sakla, WA, Sakla, AA, Chan, A & Ji, J 2010, A decision-level fusion scheme using the support vector data description for target detection in hyperspectral imagery. in Automatic Target Recognition XX; Acquisition, Tracking, Pointing, and Laser Systems Technologies XXIV; and Optical Pattern Recognition XXI. vol. 7696, 76960Y, Automatic Target Recognition XX; Acquisition, Tracking, Pointing, and Laser Systems Technologies XXIV; and Optical Pattern Recognition XXI, Orlando, FL, United States, 5/4/10. https://doi.org/10.1117/12.849936
Sakla WA, Sakla AA, Chan A, Ji J. A decision-level fusion scheme using the support vector data description for target detection in hyperspectral imagery. In Automatic Target Recognition XX; Acquisition, Tracking, Pointing, and Laser Systems Technologies XXIV; and Optical Pattern Recognition XXI. Vol. 7696. 2010. 76960Y https://doi.org/10.1117/12.849936
Sakla, Wesam A. ; Sakla, Adel A. ; Chan, Andrew ; Ji, Jim. / A decision-level fusion scheme using the support vector data description for target detection in hyperspectral imagery. Automatic Target Recognition XX; Acquisition, Tracking, Pointing, and Laser Systems Technologies XXIV; and Optical Pattern Recognition XXI. Vol. 7696 2010.
@inproceedings{bde7375bf3f74c1ebc0e62926a1a8605,
title = "A decision-level fusion scheme using the support vector data description for target detection in hyperspectral imagery",
abstract = "Spectral variability remains a challenging problem for target detection in hyperspectral (HS) imagery. In previous work, we developed a target detection scheme using the kernel-based support vector data description (SVDD). We constructed a first-order Markov-based Gaussian model to generate samples to describe the spectral variability of the target class. However, the Gaussian-generated samples also require selection of the variance parameter σ2 that dictates the level of variability in the generated target class signatures. In this work, we have investigated the use of decision-level fusion techniques for alleviating the problem of choosing a proper value of σ2. We have trained a collection of SVDDs with unique variance parameters σ2 for each of the target training sets and have investigated their combination using the traditional AND, OR, and majority vote (MV) decision-level rules. We have inserted target signatures into an urban HS scene with differing levels of spectral variability to explore the performance of the proposed scheme in these scenarios. Experiments show that the MV fusion rule is the best choice, providing relatively low false positive rates (FPR) while yielding high true positive rates (TPR). Detection results show that the proposed SVDD-based decision-level scheme using the MV fusion rule is highly accurate and yields higher true positive rates (TPR) and lower false positive rates (FPR) than the adaptive matched filter (AMF).",
keywords = "Automatic target recognition, Decision-level fusion, Hyperspectral imagery, Majority vote rule, Spectral variability, Support vector data description, Target detection",
author = "Sakla, {Wesam A.} and Sakla, {Adel A.} and Andrew Chan and Jim Ji",
year = "2010",
doi = "10.1117/12.849936",
language = "English",
isbn = "9780819481603",
volume = "7696",
booktitle = "Automatic Target Recognition XX; Acquisition, Tracking, Pointing, and Laser Systems Technologies XXIV; and Optical Pattern Recognition XXI",

}

TY - GEN

T1 - A decision-level fusion scheme using the support vector data description for target detection in hyperspectral imagery

AU - Sakla, Wesam A.

AU - Sakla, Adel A.

AU - Chan, Andrew

AU - Ji, Jim

PY - 2010

Y1 - 2010

N2 - Spectral variability remains a challenging problem for target detection in hyperspectral (HS) imagery. In previous work, we developed a target detection scheme using the kernel-based support vector data description (SVDD). We constructed a first-order Markov-based Gaussian model to generate samples to describe the spectral variability of the target class. However, the Gaussian-generated samples also require selection of the variance parameter σ2 that dictates the level of variability in the generated target class signatures. In this work, we have investigated the use of decision-level fusion techniques for alleviating the problem of choosing a proper value of σ2. We have trained a collection of SVDDs with unique variance parameters σ2 for each of the target training sets and have investigated their combination using the traditional AND, OR, and majority vote (MV) decision-level rules. We have inserted target signatures into an urban HS scene with differing levels of spectral variability to explore the performance of the proposed scheme in these scenarios. Experiments show that the MV fusion rule is the best choice, providing relatively low false positive rates (FPR) while yielding high true positive rates (TPR). Detection results show that the proposed SVDD-based decision-level scheme using the MV fusion rule is highly accurate and yields higher true positive rates (TPR) and lower false positive rates (FPR) than the adaptive matched filter (AMF).

AB - Spectral variability remains a challenging problem for target detection in hyperspectral (HS) imagery. In previous work, we developed a target detection scheme using the kernel-based support vector data description (SVDD). We constructed a first-order Markov-based Gaussian model to generate samples to describe the spectral variability of the target class. However, the Gaussian-generated samples also require selection of the variance parameter σ2 that dictates the level of variability in the generated target class signatures. In this work, we have investigated the use of decision-level fusion techniques for alleviating the problem of choosing a proper value of σ2. We have trained a collection of SVDDs with unique variance parameters σ2 for each of the target training sets and have investigated their combination using the traditional AND, OR, and majority vote (MV) decision-level rules. We have inserted target signatures into an urban HS scene with differing levels of spectral variability to explore the performance of the proposed scheme in these scenarios. Experiments show that the MV fusion rule is the best choice, providing relatively low false positive rates (FPR) while yielding high true positive rates (TPR). Detection results show that the proposed SVDD-based decision-level scheme using the MV fusion rule is highly accurate and yields higher true positive rates (TPR) and lower false positive rates (FPR) than the adaptive matched filter (AMF).

KW - Automatic target recognition

KW - Decision-level fusion

KW - Hyperspectral imagery

KW - Majority vote rule

KW - Spectral variability

KW - Support vector data description

KW - Target detection

UR - http://www.scopus.com/inward/record.url?scp=77953740054&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=77953740054&partnerID=8YFLogxK

U2 - 10.1117/12.849936

DO - 10.1117/12.849936

M3 - Conference contribution

SN - 9780819481603

VL - 7696

BT - Automatic Target Recognition XX; Acquisition, Tracking, Pointing, and Laser Systems Technologies XXIV; and Optical Pattern Recognition XXI

ER -