Noncontact Extraction of Breathing Waveform

Abstract

This paper presents a video-based method to extract breathing waveform from the upper body movement associated with breathing, which is based on Eulerian Motion Magnification and principal component analysis (PCA). Using our method, we are able to obtain the breathing waveform which contains different breathing pattern (normal breath, hold breath, quick breath, deep breath) exactly and rapidly from video without tracking the motion of body. Our method can also be used to extract the vibration pattern of other objects.