To achieve high computational efficiency and realistic visual effects, a new simulation algorithm for soft tissue deformation, which is based on a shape-matching scheme using splat primitives, is presented for interactive real-time applications, such as surgery simulation and video games. The most important novelty of the proposed approach lies in the fact that surface splats instead of points are employed in the computation of the deformation and fracturing of an elastic-plastic object. By controlling the sampling density and automatically adjusting the size of the circular splats, the surface of the simulated object can be seamlessly covered with a much small number of splats than points. Splats are then divided into clusters using the K-Means clustering algorithm. As a result, the elastic-plastic deformation of these clusters can be simulated using a shape-matching strategy, allowing more degrees of freedom (DOFs) in the simulation. Experimental results demonstrate that the proposed algorithm enormously reduces memory space and greatly improves computational efficiency (approximately twice in simulating plastic deformations compared with classical shape-matching methods), making it more suitable for interactive and real-time applications.

Additional Metadata
Keywords Degrees of freedom, Elastic-plastic model, Fracturing, K-means, Splat
Persistent URL
Journal Computers in Biology and Medicine
Zou, Y. (Yanni), & Liu, P. (2017). A new deformation simulation algorithm for elastic-plastic objects based on splat primitives. Computers in Biology and Medicine, 83, 84–93. doi:10.1016/j.compbiomed.2017.02.007