下載電子全文宣告This thesis is authorized to indicate in-campus access immediately and off-campus access at 5 years
You can not download at the moment.
Your IP address is 184.108.40.206
The defense date of the thesis is 2015-09-01
The current date is 2019-03-22
This thesis will be accessible at 2020-09-01
URN etd-0831115-174929 Statistics This thesis had been viewed 822 times. Download 3 times. Author BING-HAO HUANG Author's Email Address No Public. Department Electrical Engineering Year 2014 Semester 2 Degree Master Type of Document Master's Thesis Language English Page Count 76 Title KINECT SRNSING THE ROBOT HAND
MOTION TRAJECTORY ON TRACKING
Keyword Kinect sensor 3D hand gesture recognition hand action hand action 3D hand gesture recognition Kinect sensor Abstract In this thesis, Kinect sensor is used to build a robust 3D hand gesture recognition system in hand action tracking such that the hand of the robot in the remote site can duplicate the actions that of the human. As compared to actions of the entire human body, the hand action behaves more complex articulations and are more easily affected by kinematics modeling errors. It is thus a very attractive and challenging problem to transfer the hand gestures of the human after recognition to that of the robot. The robust hand gesture recognition is proposed to handle the kinematics and inverse kinematics from the hand wrist trajectory obtained by the Kinect sensor.
According to the hand inverse kinematics, the distance metric between human and the Kinect sensor is used to measure the dissimilarity between human and robot by hand shapes and wrist trajectory. As it only matches the wrist tip parts while not the whole hand, it can better distinguish the hand gestures of slight differences by inverse kinematics. The experiments demonstrate that the proposed hand gesture recognition system is accurate (a 99.2% mean accuracy on a challenging 5-gesture dataset), can work in distortions and orientation or scale changes, and uncontrolled environments (cluttered backgrounds and lighting conditions).
Advisor Committee Wen-Shyong Yu - advisor
none - co-chair
none - co-chair
Files Date of Defense 2015-07-30 Date of Submission 2015-09-01