Gesture recognition using dynamic time warping and kinect: A practical approach


Abstract:

This paper describes a detailed methodology that allows gesture recognition performed with the upper limbs of the human body, based on Dynamic Time Warping (DTW) technique using a Kinect in order to control a slide-show. Several involved parameters on the DTW algorithm will be analyzed allowing an accurate recognition, considering variations on the gesture velocity and even if subject's physical shape differ. A software was developed in Microsoft Visual C# to show and test how different parameters affects the performance of the algorithm. Finally an accuracy of 84.07% was obtained in this approach.

Año de publicación:

2018

Keywords:

  • Dynamic time warping
  • Human-machine interaction
  • Gesture recognition
  • kinect

Fuente:

scopusscopus

Tipo de documento:

Conference Object

Estado:

Acceso restringido

Áreas de conocimiento:

  • Visión por computadora
  • Ciencias de la computación

Áreas temáticas:

  • Física aplicada
  • Métodos informáticos especiales