ROBUST PHASE DIFFERENCE ESTIMATION OF TRANSIENTS IN HIGH NOISE LEVELS

Research output: Chapter in Book/Report/Conference proceedingPaper in conference proceedingpeer-review

Abstract

This paper presents the Reassignment Vector Phase Difference Estimator (RVPDE), which gives noise robust relative phase estimates of oscillating transient signals in high noise levels. Estimation of relative phase information between signals is of interest for direction of arrival estimation, source separation and spatio-temporal decoding in neurology as well as for soundscape analysis. The RVPDE relies on the spectrogram reassignment vectors which contains information of the time-frequency local phase difference between two transient signals. The final estimate, which is robust to high noise levels, is given as the median over the local time-frequency area. The proposed technique is shown to outperform state-of-the-art methods in simulations for high noise levels. A discussion on the statistical distribution of the estimates is also presented, and finally an example of phase difference estimation of visually evoked potentials measured from electrical brain signals is shown.

Original languageEnglish
Title of host publication30th European Signal Processing Conference, EUSIPCO 2022 - Proceedings
PublisherEuropean Signal Processing Conference, EUSIPCO
Pages2271-2275
Number of pages5
ISBN (Electronic)9789082797091
Publication statusPublished - 2022
Event30th European Signal Processing Conference, EUSIPCO 2022 - Belgrade, Serbia
Duration: 2022 Aug 292022 Sep 2

Publication series

NameEuropean Signal Processing Conference
Volume2022-August
ISSN (Print)2219-5491

Conference

Conference30th European Signal Processing Conference, EUSIPCO 2022
Country/TerritorySerbia
CityBelgrade
Period2022/08/292022/09/02

Subject classification (UKÄ)

  • Signal Processing

Keywords

  • EEG
  • low SNR
  • oscillating transient signals
  • phase estimation
  • time-frequency reassignment

Fingerprint

Dive into the research topics of 'ROBUST PHASE DIFFERENCE ESTIMATION OF TRANSIENTS IN HIGH NOISE LEVELS'. Together they form a unique fingerprint.

Cite this