Abstract
Motion capture systems have recently experienced a strong evolution. New cheap depth sensors and open source frameworks, such as OpenNI, allow for perceiving human motion on-line without using invasive systems. However, these proposals do not evaluate the validity of the obtained poses. This paper addresses this issue using a model-based pose generator to complement the OpenNI human tracker. The proposed system enforces kinematics constraints, eliminates odd poses and filters sensor noise, while learning the real dimensions of the performer's body. The system is composed by a PrimeSense sensor, an OpenNI tracker and a kinematics-based filter and has been extensively tested. Experiments show that the proposed system improves pure OpenNI results at a very low computational cost.
Highlights
There are many application fields for systems able to capture human motion
Model-free approaches have experienced a recent growing interest since the publication of the Shotton et al contributions [8]. These contributions describe the method Microsoft uses in its Kinect for Windows SDKto extract the human pose, from the depth data provided by the PrimeSense sensor
This second task is achieved by using an adaptive filter, which takes as input the set of 3D centroids provided by the OpenNI tracker, p~i
Summary
There are many application fields for systems able to capture human motion. These fields include, but are not limited to medical and rehabilitation scenarios, human-machine interfaces, computer graphics animation applied to video games or movies, surveillance, service robotics, etc. In order to avoid the previous issue and to provide a more detailed description of the performed motion, commercial marker-based HMC systems are able to capture human motion at high frame rates, which may exceed 60 frames per second (fps). They require post-processing stages to refine results and filter outliers. The first step collects 3D centroids of different body parts using the OpenNI HMC algorithm; the second step uses a model-based filter to estimate a valid human pose from these intermediate data.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.