大島商船高等専門学校紀要

Back to Top

大島商船高等専門学校紀要 Volume 45
published_at 2012-12

Unusual behavior monitoring method using MHI by wavelet preprocessing and support vector machine

ウェーブレット処理とSVMを用いたMHIによる不審動作監視
fulltext
2.51 MB
OS10045000004.pdf
Descriptions
This paper prescribes an approach for providing a method of detecting unusual human behavior from video images processed by Motion History Image (MHI) method. The video input data are pre- processed by Gabor filter or Haar wavelet transform and formulate MHI which are classified by SVM learning examples of usual and unusual behaviors. The four types characteristic features derived from MHI data are evaluated.
Creator Keywords
Motion History Image
MHI
support vector machine
SVM
wavelet
Gabor filter
Haar wavelet transform