Project : alcove
Section: Scientific Foundations
Physical properties estimation of complex objects from image sequences
In the Alcove project, we reproduce many different mechanical phenomena using our own physical simulator called Spore. Such a simulator is able to generate objects having a complex physical behavior such as human intestines for example. This has many applications especially in surgery. However parameterizing Spore 4.1 with real data coming from the observation of real human organs is essential to enhance and validate the simulation. Therefore, the main goal of this work is to estimate all these physical parameters from a real image sequence, and to use the reconstructed solution in Spore. We have started to read many recent scientific contributions regarding this problem. The conclusion is that very few works exist to compute real mechanical parameters from videos. During the first three months of this research, we have developed a new physical model for cloth animation expandable to human skin simulation, mapped on the Kawabata parameters. It is a very interesting model, because the Kawabata machine is device capable of estimating the exact physical properties of a cloth stretched by external forces. Having a good physical model being able to fully reproduce such behaviors will be very useful for our further identification of mechanical properties from images.