Measurement of sinusoidal vibration from motion blurred images

Shigang Wang, Baiqing Guan, Guobao Wang, Qian Li

Research output: Contribution to journalArticle

31 Scopus citations

Abstract

Previous vision-based methods usually measure vibration from large sequence of unblurred images recorded by high-speed video or stroboscopic photography. In this paper, we propose a novel method for sinusoidal vibration measurement based on motion blurred images. We represent the motion blur information in images by the relationship between the geometric moments of the motion blurred images and the motion, and estimate the vibration parameters from this motion blur cue. We need only one motion blurred image and an unblurred image or two successive frames of blurred images to calculate the parameters of low-frequency vibration as well as the amplitude and direction of high-frequency vibration, while unblurred-image-based techniques rely on much more images to obtain the same results and existing motion-blurred-image-based approaches only estimate the amplitude of high-frequency vibration. Experimental results with both simulated and real vibrations of low and high frequencies are employed to demonstrate the effectiveness of the proposed scheme.

Original languageEnglish (US)
Pages (from-to)1029-1040
Number of pages12
JournalPattern Recognition Letters
Volume28
Issue number9
DOIs
Publication statusPublished - Jul 1 2007
Externally publishedYes

    Fingerprint

Keywords

  • Geometric moment
  • Motion blur
  • Motion blurred image
  • Vibration measurement

ASJC Scopus subject areas

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

Cite this