TY - JOUR
T1 - Implementation of an omnidirectional human motion capture system using multiple Kinect sensors
AU - Kim, Junghwan
AU - Lee, Inwoong
AU - Kim, Jongyoo
AU - Lee, Sanghoon
N1 - Publisher Copyright:
Copyright © 2015 The Institute of Electronics, Information and Communication Engineers.
Copyright:
Copyright 2017 Elsevier B.V., All rights reserved.
PY - 2015/9/1
Y1 - 2015/9/1
N2 - Due to ease of implementation for various user interactive applications, much research on motion recognition has been completed using Kinect. However, one drawback of Kinect is that the skeletal information obtained is provided under the assumption that the user faces Kinect. Thus, the skeletal information is likely incorrect when the user turns his back to Kinect, which may lead to difficulty in motion recognition from the application. In this paper, we implement a highly accurate human motion capture system by installing six Kinect sensors over 360 degrees. The proposed method enables skeleton to be obtained more accurately by assigning higher weights to skeletons captured by Kinect in which the user faces forward. Toward this goal, the front vector of the user is temporally traced to determine whether the user is facing Kinect. Then, more reliable joint information is utilized to construct a skeletal representation of each user.
AB - Due to ease of implementation for various user interactive applications, much research on motion recognition has been completed using Kinect. However, one drawback of Kinect is that the skeletal information obtained is provided under the assumption that the user faces Kinect. Thus, the skeletal information is likely incorrect when the user turns his back to Kinect, which may lead to difficulty in motion recognition from the application. In this paper, we implement a highly accurate human motion capture system by installing six Kinect sensors over 360 degrees. The proposed method enables skeleton to be obtained more accurately by assigning higher weights to skeletons captured by Kinect in which the user faces forward. Toward this goal, the front vector of the user is temporally traced to determine whether the user is facing Kinect. Then, more reliable joint information is utilized to construct a skeletal representation of each user.
UR - http://www.scopus.com/inward/record.url?scp=84940974051&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84940974051&partnerID=8YFLogxK
U2 - 10.1587/transfun.E98.A.2004
DO - 10.1587/transfun.E98.A.2004
M3 - Article
AN - SCOPUS:84940974051
VL - E98A
SP - 2004
EP - 2008
JO - IEICE Transactions on Fundamentals of Electronics, Communications and Computer Sciences
JF - IEICE Transactions on Fundamentals of Electronics, Communications and Computer Sciences
SN - 0916-8508
IS - 9
ER -