Real-Time Markerless Tracking the Human Hands for 3D Interaction

Dissertation, Universität Bonn, Feb. 2011
 

Abstract

This thesis presents methods for enabling suitable human computer interaction using only movements of the bare human hands in free space. This kind of interaction is natural and intuitive, particularly because actions familiar to our everyday life can be reflected. Furthermore, the input is contact-free which is of great advantage e.g. in medical applications due to hygiene factors.

For enabling the translation of hand movements to control signals an automatic method for tracking the pose and/or posture of the hand is needed. In this context the simultaneous recognition of both hands is desirable to allow for more natural input. The first contribution of this thesis is a novel video-based method for real-time detection of the positions and orientations of both bare human hands in four different predefined postures, respectively. Based on such a system novel interaction interfaces can be developed. However, the design of such interfaces is a non-trivial task. Additionally, the development of novel interaction techniques is often mandatory in order to enable the design of efficient and easily operable interfaces. To this end, several novel interaction techniques are presented and investigated in this thesis, which solve existing problems and substantially improve the applicability of such a new device. These techniques are not restricted to this input instrument and can also be employed to improve the handling of other interaction devices. Finally, several new interaction interfaces are described and analyzed to demonstrate possible applications in specific interaction scenarios.

(Thesis submission date: Apr. 2010)

Download: http://nbn-resolving.de/urn:nbn:de:hbz:5N-24232

Bibtex

@PHDTHESIS{schlattmann-2011-dissertation,
    author = {Schlattmann, Markus},
     title = {Real-Time Markerless Tracking the Human Hands for 3D Interaction},
      type = {Dissertation},
      year = {2011},
     month = feb,
    school = {Universit{\"a}t Bonn},
  abstract = {This thesis presents methods for enabling suitable human computer interaction using only movements
              of the bare human hands in free space. This kind of interaction is natural and intuitive,
              particularly because actions familiar to our everyday life can be reflected. Furthermore, the input
              is contact-free which is of great advantage e.g. in medical applications due to hygiene factors.
              
              For enabling the translation of hand movements to control signals an automatic method for tracking
              the pose and/or posture of the hand is needed. In this context the simultaneous recognition of both
              hands is desirable to allow for more natural input. The first contribution of this thesis is a novel
              video-based method for real-time detection of the positions and orientations of both bare human
              hands in four different predefined postures, respectively. Based on such a system novel interaction
              interfaces can be developed. However, the design of such interfaces is a non-trivial task.
              Additionally, the development of novel interaction techniques is often mandatory in order to enable
              the design of efficient and easily operable interfaces. To this end, several novel interaction
              techniques are presented and investigated in this thesis, which solve existing problems and
              substantially improve the applicability of such a new device. These techniques are not restricted to
              this input instrument and can also be employed to improve the handling of other interaction devices.
              Finally, several new interaction interfaces are described and analyzed to demonstrate possible
              applications in specific interaction scenarios.},
       url = {http://nbn-resolving.de/urn:nbn:de:hbz:5N-24232}
}