Higher-Order Spectral Analysis of Human Motion

We describe a higher-orderspectralanalysis-basedapproach for detecting people by recognizing human motion such as walking or running. The periodic attribute of human motion lends itself to efficient spectral inspection. In the proposed method, the stridelengthis determined in every frame asthe image sequenceevolves. The bispectrum which is the Fourier transform of the triple correlation is a robust indicator of presence of periodicity. Triple correlation is robust as it is immune to any symmetricallydistributed noise. The method is successfully tested on real video sequences.

• Type: Conference
• Year: 2000
• Where: ICIP
• Authors: A. N. Rajagopalan, Rama Chellappa
• Keywords: Efficient Spectral Inspection, Human Motion, ICIP 2000, Image Processing, Triple Correlation