GPU Accelerated Particle System for Triangulated Surface Meshes

Shape analysis based on images and implicit surfaces has been an active area of research for the past several years. Particle systems have emerged as a viable solution to represent shapes for statistical analysis. One of the most widely used representations of shapes in computer graphics and visualization is the triangular mesh. It is desirable to provide a particle system representation for statistical analysis of meshes. This paper presents a framework to distribute particles over a surface defined by a triangle mesh and provides an efficient and controllable mechanism to sample shapes defined by triangular meshes within a locally adaptive scheme. We propose GPU implementations to leverage inherent data parallelism. Results on triangular meshes representing synthetic and real shapes demonstrate speedups on the GPU upto 341X as compared to the CPU implementation.