3D Reconstruction of Human Motion from Video

In proceedings of Workshop on Games and Graphics 2013, Dez. 2013
 

Abstract

This paper presents a novel framework for 3D full body reconstruction of human motion from uncalibrated monocular video data. We rst detect and track feature sets from video sequences by employing MSER and SURF feature detection techniques together with prior information obtained from the motion capture database. By deriving suitable feature sets from both video and motion capture data, we are able to employ fast motion retrieval techniques to nd the best relevant prior poses. The resulting 3D motion sequences are reconstructed by an energy minimization process that takes multiple prior terms into account.

Additional Material can be found on the papers project page: http://cg.cs.uni-bonn.de/vmotrec

Bilder

Paper herunterladen

Paper herunterladen

Bibtex

@INPROCEEDINGS{yasin-2013b,
     author = {Yasin, Hashim and Kr{\"u}ger, Bj{\"o}rn and Weber, Andreas},
      title = {3D Reconstruction of Human Motion from Video},
  booktitle = {Workshop on Games and Graphics 2013},
       year = {2013},
      month = dec,
   abstract = {This paper presents a novel framework for 3D full body reconstruction
               of human motion from uncalibrated monocular video data.
               We rst detect and track feature sets from video sequences by employing
               MSER and SURF feature detection techniques together with prior information
               obtained from the motion capture database. By deriving suitable
               feature sets from both video and motion capture data, we are able to
               employ fast motion retrieval techniques to nd the best relevant prior
               poses. The resulting 3D motion sequences are reconstructed by an energy
               minimization process that takes multiple prior terms into account.}
}