Wavelet-based feature extraction using probabilistic finite state automata for pattern classification

Xin Jin, Shalabh Gupta, Kushal Mukherjee, Asok Ray

Research output: Contribution to journalArticle

44 Citations (Scopus)

Abstract

Real-time data-driven pattern classification requires extraction of relevant features from the observed time series as low-dimensional and yet information-rich representations of the underlying dynamics. These low-dimensional features facilitate in situ decision-making in diverse applications, such as computer vision, structural health monitoring, and robotics. Wavelet transforms of time series have been widely used for feature extraction owing to their timefrequency localization properties. In this regard, this paper presents a symbolic dynamics-based method to model surface images, generated by wavelet coefficients in the scale-shift space. These symbolic dynamics-based models (e.g., probabilistic finite state automata (PFSA)) capture the relevant information, embedded in the sensor data, from the associated Perron-Frobenius operators (i.e., the state-transition probability matrices). The proposed method of pattern classification has been experimentally validated on laboratory apparatuses for two different applications: (i) early detection of evolving damage in polycrystalline alloy structures, and (ii) classification of mobile robots and their motion profiles.

Original languageEnglish (US)
Pages (from-to)1343-1356
Number of pages14
JournalPattern Recognition
Volume44
Issue number7
DOIs
StatePublished - Jul 1 2011

Fingerprint

Finite automata
Pattern recognition
Feature extraction
Time series
Structural health monitoring
Mobile robots
Wavelet transforms
Computer vision
Mathematical operators
Robotics
Decision making
Sensors

All Science Journal Classification (ASJC) codes

  • Software
  • Signal Processing
  • Computer Vision and Pattern Recognition
  • Artificial Intelligence

Cite this

Jin, Xin ; Gupta, Shalabh ; Mukherjee, Kushal ; Ray, Asok. / Wavelet-based feature extraction using probabilistic finite state automata for pattern classification. In: Pattern Recognition. 2011 ; Vol. 44, No. 7. pp. 1343-1356.
@article{ca5cd1fda2c845b596dbcf8eae05b28c,
title = "Wavelet-based feature extraction using probabilistic finite state automata for pattern classification",
abstract = "Real-time data-driven pattern classification requires extraction of relevant features from the observed time series as low-dimensional and yet information-rich representations of the underlying dynamics. These low-dimensional features facilitate in situ decision-making in diverse applications, such as computer vision, structural health monitoring, and robotics. Wavelet transforms of time series have been widely used for feature extraction owing to their timefrequency localization properties. In this regard, this paper presents a symbolic dynamics-based method to model surface images, generated by wavelet coefficients in the scale-shift space. These symbolic dynamics-based models (e.g., probabilistic finite state automata (PFSA)) capture the relevant information, embedded in the sensor data, from the associated Perron-Frobenius operators (i.e., the state-transition probability matrices). The proposed method of pattern classification has been experimentally validated on laboratory apparatuses for two different applications: (i) early detection of evolving damage in polycrystalline alloy structures, and (ii) classification of mobile robots and their motion profiles.",
author = "Xin Jin and Shalabh Gupta and Kushal Mukherjee and Asok Ray",
year = "2011",
month = "7",
day = "1",
doi = "10.1016/j.patcog.2010.12.003",
language = "English (US)",
volume = "44",
pages = "1343--1356",
journal = "Pattern Recognition",
issn = "0031-3203",
publisher = "Elsevier Limited",
number = "7",

}

Wavelet-based feature extraction using probabilistic finite state automata for pattern classification. / Jin, Xin; Gupta, Shalabh; Mukherjee, Kushal; Ray, Asok.

In: Pattern Recognition, Vol. 44, No. 7, 01.07.2011, p. 1343-1356.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Wavelet-based feature extraction using probabilistic finite state automata for pattern classification

AU - Jin, Xin

AU - Gupta, Shalabh

AU - Mukherjee, Kushal

AU - Ray, Asok

PY - 2011/7/1

Y1 - 2011/7/1

N2 - Real-time data-driven pattern classification requires extraction of relevant features from the observed time series as low-dimensional and yet information-rich representations of the underlying dynamics. These low-dimensional features facilitate in situ decision-making in diverse applications, such as computer vision, structural health monitoring, and robotics. Wavelet transforms of time series have been widely used for feature extraction owing to their timefrequency localization properties. In this regard, this paper presents a symbolic dynamics-based method to model surface images, generated by wavelet coefficients in the scale-shift space. These symbolic dynamics-based models (e.g., probabilistic finite state automata (PFSA)) capture the relevant information, embedded in the sensor data, from the associated Perron-Frobenius operators (i.e., the state-transition probability matrices). The proposed method of pattern classification has been experimentally validated on laboratory apparatuses for two different applications: (i) early detection of evolving damage in polycrystalline alloy structures, and (ii) classification of mobile robots and their motion profiles.

AB - Real-time data-driven pattern classification requires extraction of relevant features from the observed time series as low-dimensional and yet information-rich representations of the underlying dynamics. These low-dimensional features facilitate in situ decision-making in diverse applications, such as computer vision, structural health monitoring, and robotics. Wavelet transforms of time series have been widely used for feature extraction owing to their timefrequency localization properties. In this regard, this paper presents a symbolic dynamics-based method to model surface images, generated by wavelet coefficients in the scale-shift space. These symbolic dynamics-based models (e.g., probabilistic finite state automata (PFSA)) capture the relevant information, embedded in the sensor data, from the associated Perron-Frobenius operators (i.e., the state-transition probability matrices). The proposed method of pattern classification has been experimentally validated on laboratory apparatuses for two different applications: (i) early detection of evolving damage in polycrystalline alloy structures, and (ii) classification of mobile robots and their motion profiles.

UR - http://www.scopus.com/inward/record.url?scp=79952184225&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=79952184225&partnerID=8YFLogxK

U2 - 10.1016/j.patcog.2010.12.003

DO - 10.1016/j.patcog.2010.12.003

M3 - Article

AN - SCOPUS:79952184225

VL - 44

SP - 1343

EP - 1356

JO - Pattern Recognition

JF - Pattern Recognition

SN - 0031-3203

IS - 7

ER -