A Model-Based Articulated Human Motion Tracking System

[[abstract]]This paper proposes a model-based human motion analysis method using two cameras without any markers on the human actor. Using the background subtraction, we extract the silhouette of the human object from two orthogonal views. To track the human actor, we find the best match between the articulated human model and the binary human object. We propose a 3-level estimation algorithm which consists of low-level analysis, mid-level estimation and high-level verification and correction. Moreover, to relax the limitations of using the articulated human model, we predefine some auxiliary postures, and develop a projection profile analysis to recognize the postures of bending and rotating. Finally, to evaluate the performance of the tracking system, we use HMMs to recognize different human motions.[[fileno]]2030109030110[[department]]電機工程學