Occlusion-adaptive fusion for gait-based motion recognition
Shiloh L. Dockstader · 2003
This paper presents a new architecture for motion- and video-based event recognition using the fusion of multiple hidden Markov models (HMw with a Bayesian belief network. We begin with a fifteen para- meter structural model of the human body, where unique parameter groups and extracted gait variables define individual nodes of the network. Each node is character- ized by a conditional probability mass function (PMF) in addition to evidence regarding its current state given a set of observations. The evidence for each state is virtual and derived from the conditional ou@ut probabilities of two HMM; one represents a fundamental activity while the other defines a tracking failure event. This novel inte- gration provides a means of recognizing a variety of activities in the presence of noise, occlusion, ambiguity, and entirely missing observations. We demonstrate the effectiveness of our approach on numerous multi-view video sequences of complex human motion.