Estimation of Occluded Lower Limb Using Inverted Pendulum Model
For many interactive vision based systems, users are captured by the system and modeled as a simplified skeleton. The skeleton model usually encounter problems of occlusion because of obstacles or other users appearing in the environment. In this paper, we proposed a method based on an inverted pendulum model (IPM), which was applied to recover the occluded leg of skeleton model. The skeleton model was provided by Vicon motion measurement system, which captured the participant's motions. One leg of the skeleton model was removed intentionally and the proposed method was used to estimate the pose of the occluded leg in real-time. The results showed that most of the positioning errors were within 10 cm on average and the processing rate exceeded 100 fps.