RECOGNIZING DIFFERENT TYPES OF STOCHASTIC PROCESSES Academic Article

abstract

• We propose a new cross-correlation method that can recognize independent realizations of the same type of stochastic processes and can be used as a new kind of pattern recognition tool in biometrics, sensing, forensic, security and image processing applications. The method, which we call bispectrum correlation coefficient method, makes use of the cross-correlation of the bispectra. Three kinds of cross-correlation coefficients are introduced. To demonstrate the new method, six different random telegraph signals are tested, where four of them have the same power density spectrum. It is shown that the three coefficients can map the different stochastic processes to specific sub-volumes in a cube.

authors

• Kish, Laszlo

published proceedings

• FLUCTUATION AND NOISE LETTERS

author list (cited authors)

• Kim, J. U., & Kish, L. B.

citation count

• 1

complete list of authors

• Kim, JU||Kish, LB

publication date

• March 2006

publisher

• World Scientific Publishing  Publisher

published in

• Fluctuation and Noise Letters (FNL)  Journal

keywords

• Biometrics
• Bispectra
• Forensic
• Pattern Recognition
• Security
• Stochastic Process Recognition

Digital Object Identifier (DOI)

• 10.1142/S0219477506003082

start page

• L1

end page

• L6

volume

• 6

issue

• 1

URL

• http://dx.doi.org/10.1142/s0219477506003082