3D Modeling, Processing and Transmission
The LTTM (multimedia telecommunication) group research activity is focused on construction and transmission of 3D data.
Current research areas include markerless motion capture (3D video), transmission and remote visualization of 3D scenes, automatic generation of 3D representations and content-based retrieval of 3D data.
More informations can be found in the research group website http://lttm.dei.unipd.it
Markerless Motion Capture
Markerless Human Motion Capture Project aims to design and to develop a system for capturing the human motion using passive video cameras for free-viewpoint video purposes. The data is acquired in a 24 square meters room furnished with blue walls using four high resolution cameras. Actors in the scene are represented as articulated deformable models and their deformations are modeled according to the Skeletal Sub-Space Deformation (SSD) framework. We developed a new approach to marker-less motion tracking of the skinned meshes using both optical
flow and silhouette information extracted from the four videos.
Transmission and remote visualization of 3D scenes
We are currently working in collaboration with the University of New South Wales (Australia) to a new client-server paradigm for the transmission of textured 3D models. Research activity is focused on the optimal allocation of the available transmission resources between geometry and texture data in a progressive and scalable compression and transmission scheme. To achieve this target an extension of the JPIP standard and communication scheme for JPEG2000 images to 3D scene browsing is currently under development. Furthermore a joint research activity with the SIGNET group concerns the study of 3D transmission over lossy channels.