Parallel Surface Reconstruction on GPU
State Key Laboratory for Novel Software Technology, Department of Computer Science And Technology, Nanjing University Nanjing, China
International Conference on Internet Multimedia Computing and Service (ICIMCS’15), China
@article{gao2015parallel,
title={Parallel Surface Reconstruction on GPU},
author={Gao, Heng and Tang, Jie and Wu, Gangshan},
year={2015}
}
Marching Cubes is the most frequently used method to reconstruct isosurface from a point cloud. However, the point clouds are getting denser and denser, thus the efficiency of Marching cubes method has become an obstacle. This paper presents a novel GPU-based parallel surface reconstruction algorithm. The algorithm firstly creates a GPU-based uniform grid structure to manage point cloud. Then directed distances from vertices of cubes to the point cloud are computed in a newly put forwarded parallel way. Finally, after the generation of triangles, a space indexing scheme is adopted to reconstruct the connectivity of the resulted surface. The results show that our algorithm can run more than 10 times faster compared to the CPU-based implementations.
August 1, 2015 by hgpu
