• Home
• News & Events
• Publications
• Projects
• Thesis and PhD
• People
• Links

• Sens
• Beauty
• Kin
• E-ViP
• HMouse
• MCap
• MRec
• VHull
• VDrape

Main Content

Full Body Motion Capture

Our approach is based on multiple 2D silhouettes of the body extracted from 2D images and can be outlined as follows:

• different cameras are used to obtain views of a human body. From each image a 2D silhouette of the performer is extracted
• a volumetric description of the object is recovered by intersecting the solid cones obtained by back projecting from each viewpoint the corresponding silhouette (Volume Intersection, VI). The final voxel representation can be obtained at different resolutions
• the posture is recovered by fitting a model of the human body to the reconstructed volume. This is obtained by minimizing a suitable distance function between the volume and the model with a search through the space of pose parameters

Head & Facial Features Tracking

We have also developed a system to track head position and facial features in real time. First the head position is reconstructed by means of a model based head tracker, which matches a 3D generic textured head model and 2D image features extracted from the input sequence. Head tracking is achieved by minimizing an error function which describes the discrepancies between model and image features. Motion and texture information are considered in order to make tracking robust.

After pose reconstruction, the input image is warped into the texture map of the model. The resulting stabilized view of the face is then processed to reconstruct significant face features, like eyes, brows and lips. The proposed method uses multi-state deformable templates which are fitted to the warped image exploiting motion, shape and color information. The final output of the system is a stream of head and face animation parameters. The overall performance of the head tracking algorithm is about 30 frame/s on a Pentium III 600, while face expression reconstruction process runs at about 45 frame/s

Related Papers

• A. Bottino, A. Laurentini
  A Silhouette-based technique for the Reconstruction of Human Movement
  Computer Vision and Image Understanding, Vol. 83, No. 1, pp.79-95, 2001 (PDF)

• A. Bottino, A. Laurentini
  Experimenting with Motion Capture in a Virtual Environment
  The Visual Computer Journal, Vol. 17, No. 1, 2001 (PDF) (DOI)

• A. Bottino
  Real time head and facial features tracking from uncalibrated monocular views
  Proceedings of 5th Asian Conference on Computer Vision (ACCV 2002), Melbourne, Australia January 23-25, 2002 (PDF)

• A. Bottino
  Real time head tracking from uncalibrated monocular views
  Proceedings of 5th World Multiconference on Systemics, Cybernetics and Informatics (SCI 2001), Orlando (FL), 22-25 July, 2001 (PDF)

• A. Bottino, A. Laurentini
  Non-Intrusive Silhouette based Motion Capture
  Proceedings of 4th World Multiconference on Systemics, Cybernetics and Informatics SCI 2000, Orlando (FL), 23-26 July 2000 (PDF) Selected as best paper presented in the session Image and Multidimensional Signal Processing

• A. Bottino, A.Laurentini, P. Zuccone,
  Toward Non-intrusive Motion Capture ,
  proceedings of Asian Conference on Computer Vision '98 (ACCV'98), Hong Kong, 8-11 January 98 (PS.ZIP)