Motion composition of 3d video

  • Authors:
  • Jianfeng Xu;Toshihiko Yamasaki;Kiyoharu Aizawa

  • Affiliations:
  • Dept. of Electronics Engineering;Dept. of Frontier Informatics, The University of Tokyo, Chiba, Japan;Dept. of Electronics Engineering

  • Venue:
  • PCM'06 Proceedings of the 7th Pacific Rim conference on Advances in Multimedia Information Processing
  • Year:
  • 2006

Quantified Score

Hi-index 0.00

Visualization

Abstract

3D video, which is composed of a sequence of mesh models and can provide the user with interactivity, is attracting increasing attention in many research groups. However, it is time-consuming and expensive to generate 3D video sequences. In this paper, a motion composition method is proposed to edit 3D video based on the user’s requirements so that 3D video can be re-used. By analyzing the feature vectors, the hierarchical motion structure is parsed and then a motion database is set up by selecting the representative motions. A motion graph is constructed to organize the motion database by finding the possible motion transitions. Then, the best path is searched based on a proposed cost function by a modified Dijkstra algorithm after the user selects the desired motions in the motion database, which are called key motions in this paper. Our experimental results show the edited 3D video sequence looks natural and realistic.