Fast Sparse Level Sets on Graphics Hardware.

Area of Science:

• Computer Graphics
• Scientific Visualization
• Computational Geometry

Background:

• Level-set methods are widely used for tracking deformable interfaces in computer graphics.
• Interactive simulations using level sets are computationally demanding, limiting their application.
• Existing parallel approaches on CPU and GPU hardware have performance limitations.

Purpose of the Study:

• To develop a computationally efficient sparse GPU level-set method.
• To accelerate interactive simulations and surface reconstruction using graphics processors.
• To enable real-time performance for level-set computations and rendering.

Main Methods:

• Leveraging graphics processing units (GPUs) for accelerated level-set computations.
• Implementing an efficient, sparse GPU level-set algorithm.
• Developing a novel multiresolution method for surface reconstruction from point clouds using GPU level sets.

Main Results:

• The sparse GPU level-set method achieves substantially higher speeds compared to state-of-the-art parallel methods on both CPU and GPU.
• The novel surface reconstruction method demonstrates favorable performance against existing techniques and parallel implementations.
• Both level-set computations and rendering of level-set surfaces are performed at interactive rates, even on large volumetric grids.

Conclusions:

• The developed sparse GPU level-set method significantly overcomes the computational challenges of traditional level-set techniques.
• This approach enables interactive rates for complex simulations and surface reconstruction tasks.
• Numerous applications leveraging level sets can benefit from the proposed efficient sparse GPU method for enhanced performance.