Modelling of elastic deformation using stereo vision and smoothed particle hydrodynamics

Elastic surface deformation of an object is described using depth information obtained at video frame-rate with camera calibration and stereo matching techniques. Two synchronised USB cameras acquire successive image pairs describing the surface deformation. Images are corrected to exclude lens distortion and generate rectified stereo pairs. The depth information is computed with a simplified version of the belief propagation stereo matching algorithm. Experimental results are related to computer simulation of elastic deformation based on the mesh-free smoothed particle hydrodynamics technique. The obtained strain, stress, and internal energy of the object help calibrate the physical properties of the computational model and lead to more realistic simulation of elastic deformation.