Selective elasticity data acquisition on 3D deformable objects for virtualized reality applications

This paper proposes the use of self-organizing architectures, particularly the growing neural gas, for the purpose of automatically guiding elasticity data acquisition based on a sparse vision and elasticity point-cloud of a 3D object. The proposed solution allows for the identification of regions where changes in the elastic behavior of the object occur. Additional data can then be collected in these areas in order to better characterize the elastic characteristics of a certain object. Experimental results for different non-homogeneous objects are presented in order to validate the proposed solution.