Recognition using information-optimal adaptive feature-specific imaging

Pawan K. Baheti, Mark A Neifeld

Research output: Contribution to journalArticle

31 Citations (Scopus)

Abstract

We present an information-theoretic adaptive feature-specific imaging (AFSI) system for a M-class recognition task. The proposed system utilizes the recently developed task-specific information (TSI) framework to incorporate he knowledge from previous measurements and adapt the projection matrix at each step. The decisionmaking ramework is based on sequential hypothesis testing. We quantify the number of measurements required to achieve a specified probability of misclassification (Pe), and we compare the performances of three approaches: the new TSI-based AFSI system, a previously reported statistical AFSI system, and static FSI (SFSI). The TSI-based AFSI system exhibits significant improvement compared with SFSI and statistical AFSI at low signal-to-noise ratio (SNR). It is shown that for M=4 hypotheses, SNR=-20 dB and desired Pe=10-2, TSI-based AFSI requires 3 times fewer measurements than statistical AFSI, and 16 times fewer measurements than SFSI. We also describe an extension of the proposed method that is suitable for recognition in the presence of nuisance parameters such as illumination conditions and target orientations.

Original languageEnglish (US)
Pages (from-to)1055-1070
Number of pages16
JournalJournal of the Optical Society of America A: Optics and Image Science, and Vision
Volume26
Issue number4
DOIs
StatePublished - Apr 1 2009

Fingerprint

Signal-To-Noise Ratio
Imaging systems
Imaging techniques
Time measurement
Lighting
Signal to noise ratio
signal to noise ratios
time measurement
projection
illumination
Testing
matrices

ASJC Scopus subject areas

  • Atomic and Molecular Physics, and Optics
  • Electronic, Optical and Magnetic Materials
  • Computer Vision and Pattern Recognition

Cite this

Recognition using information-optimal adaptive feature-specific imaging. / Baheti, Pawan K.; Neifeld, Mark A.

In: Journal of the Optical Society of America A: Optics and Image Science, and Vision, Vol. 26, No. 4, 01.04.2009, p. 1055-1070.

Research output: Contribution to journalArticle

@article{a8459531271d4be8b60b9b6c1c1c65be,
title = "Recognition using information-optimal adaptive feature-specific imaging",
abstract = "We present an information-theoretic adaptive feature-specific imaging (AFSI) system for a M-class recognition task. The proposed system utilizes the recently developed task-specific information (TSI) framework to incorporate he knowledge from previous measurements and adapt the projection matrix at each step. The decisionmaking ramework is based on sequential hypothesis testing. We quantify the number of measurements required to achieve a specified probability of misclassification (Pe), and we compare the performances of three approaches: the new TSI-based AFSI system, a previously reported statistical AFSI system, and static FSI (SFSI). The TSI-based AFSI system exhibits significant improvement compared with SFSI and statistical AFSI at low signal-to-noise ratio (SNR). It is shown that for M=4 hypotheses, SNR=-20 dB and desired Pe=10-2, TSI-based AFSI requires 3 times fewer measurements than statistical AFSI, and 16 times fewer measurements than SFSI. We also describe an extension of the proposed method that is suitable for recognition in the presence of nuisance parameters such as illumination conditions and target orientations.",
author = "Baheti, {Pawan K.} and Neifeld, {Mark A}",
year = "2009",
month = "4",
day = "1",
doi = "10.1364/JOSAA.26.001055",
language = "English (US)",
volume = "26",
pages = "1055--1070",
journal = "Journal of the Optical Society of America A: Optics and Image Science, and Vision",
issn = "1084-7529",
publisher = "The Optical Society",
number = "4",

}

TY - JOUR

T1 - Recognition using information-optimal adaptive feature-specific imaging

AU - Baheti, Pawan K.

AU - Neifeld, Mark A

PY - 2009/4/1

Y1 - 2009/4/1

N2 - We present an information-theoretic adaptive feature-specific imaging (AFSI) system for a M-class recognition task. The proposed system utilizes the recently developed task-specific information (TSI) framework to incorporate he knowledge from previous measurements and adapt the projection matrix at each step. The decisionmaking ramework is based on sequential hypothesis testing. We quantify the number of measurements required to achieve a specified probability of misclassification (Pe), and we compare the performances of three approaches: the new TSI-based AFSI system, a previously reported statistical AFSI system, and static FSI (SFSI). The TSI-based AFSI system exhibits significant improvement compared with SFSI and statistical AFSI at low signal-to-noise ratio (SNR). It is shown that for M=4 hypotheses, SNR=-20 dB and desired Pe=10-2, TSI-based AFSI requires 3 times fewer measurements than statistical AFSI, and 16 times fewer measurements than SFSI. We also describe an extension of the proposed method that is suitable for recognition in the presence of nuisance parameters such as illumination conditions and target orientations.

AB - We present an information-theoretic adaptive feature-specific imaging (AFSI) system for a M-class recognition task. The proposed system utilizes the recently developed task-specific information (TSI) framework to incorporate he knowledge from previous measurements and adapt the projection matrix at each step. The decisionmaking ramework is based on sequential hypothesis testing. We quantify the number of measurements required to achieve a specified probability of misclassification (Pe), and we compare the performances of three approaches: the new TSI-based AFSI system, a previously reported statistical AFSI system, and static FSI (SFSI). The TSI-based AFSI system exhibits significant improvement compared with SFSI and statistical AFSI at low signal-to-noise ratio (SNR). It is shown that for M=4 hypotheses, SNR=-20 dB and desired Pe=10-2, TSI-based AFSI requires 3 times fewer measurements than statistical AFSI, and 16 times fewer measurements than SFSI. We also describe an extension of the proposed method that is suitable for recognition in the presence of nuisance parameters such as illumination conditions and target orientations.

UR - http://www.scopus.com/inward/record.url?scp=65249109798&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=65249109798&partnerID=8YFLogxK

U2 - 10.1364/JOSAA.26.001055

DO - 10.1364/JOSAA.26.001055

M3 - Article

C2 - 19340282

AN - SCOPUS:65249109798

VL - 26

SP - 1055

EP - 1070

JO - Journal of the Optical Society of America A: Optics and Image Science, and Vision

JF - Journal of the Optical Society of America A: Optics and Image Science, and Vision

SN - 1084-7529

IS - 4

ER -