Section: New Results

Keywords : 3D interaction, augmented reality, mediated reality, mixed reality.

Virtual Reality

Participants : Jean-Marc Hasenfratz, Jean-dominique Gascuel, Marc Lapierre, Alexandrina Orzan.

CYBER II: Omnidirectional texturing

To make the model in the CYBER II framework (see section  3.5.1 ) more realistic, we map the images captured from the video streams onto the 3D shape. Since we have a set of cameras, we have more then one view for the same patch of the 3D model. In addition, the geometry of the model is limited to about 5000 triangles, not sufficient to have a good mesh representation, but necessary if we want to obtain it in real time.

In this context, we proposes to identify the problems of multi-view texturing for approaches where the 3D model is given; model-free and to offer solutions for a number of them. We successfully treat visibility issues, identify ``crisk zones'', correct projection displacement errors and fill in small untextured areas. Our algorithm works in real time, thus permitting an interactive viewing of the augmented scene. The pipline of the treatment is proposed in Figure 17 . First result are published in [21] .

Figure 17. Pipeline for multi-view texturing.

Virtual Reality and Rehabilitation

Since 2001, a collaboration was established between O. Martin (MCU UJF/UFRAPS/SPM Grenoble), C. Prablanc (DR INSERM U534 Lyon) and ARTIS. This collaboration had two symetric sides: to assess the usability of virtual reality environment for neuro-sciences experiments on the perception-action loop; and to find measures of comfort and immersion of a virtual setup. Last year, we have been joined by Paul Pavan (Kinesitherapist at the Grenoble University Hospital). He routinely uses a projection device (that generate moving light dots) to stress balance of patients, in order to do rehabilitation eg. after brain traumas.

Figure 18. The new simpler experimental environment. (left) The instrumented patient, with devices to record 3D positions, and muscular activations. (right) The patient reacting at the computer generated motion field.

We provided a flexible stimulation environment suitable to reproduce qualibrated motion field sequences, using a simple VR setup, a laptop, a video projector, and a few input devices. The aim is to record balance control and adaptation when shown various types and directions of motion fields.

A test experiment took place in june 2006 (see figure 18 . A small sample of well being people (10 ``naive'' students from another lab) where used in a first scenario. We are still exploting the numberous records, but we are already confident that the setup was indeed able to induce standard balance control mechanisms a user will have in real life world.