Ballistic hand movements

  • Authors:

  • V. Shiv Naga Prasad;Vili Kellokumpu;Larry S. Davis

  • Affiliations:

  • Perceptual Interfaces and Reality Laboratory, Univ. of Maryland – College Park, College Park, MD;Machine Vision Group, Univ. of Oulu, Oulu, Finland;Perceptual Interfaces and Reality Laboratory, Univ. of Maryland – College Park, College Park, MD

  • Venue:
  • AMDO'06 Proceedings of the 4th international conference on Articulated Motion and Deformable Objects
  • Year:
  • 2006

Quantified Score

Hi-index 0.00

Visualization

Abstract

Common movements like reaching, striking, etc. observed during surveillance have highly variable target locations. This puts appearance-based techniques at a disadvantage for modelling and recognizing them. Psychological studies indicate that these actions are ballistic in nature. Their trajectories have simple structures and are determined to a great degree by the starting and ending positions. We present an approach for movement recognition that explicitly considers their ballistic nature. This enables the decoupling of recognition from the movement's trajectory, allowing generalization over a range of target-positions. A given movement is first analyzed to determine if it is ballistic. Ballistic movements are further classified into reaching, striking, etc. The proposed approach was tested with motion capture data obtained from the CMU MoCap database