In this paper we describe a parallel implicit method based on radial basisfunctions (RBF) for surface reconstruction. The applicability of RBF methods ishindered by its computational demand, that requires the solution of linearsystems of size equal to the number of data points. Our reconstructionimplementation relies on parallel scientific libraries and is supported formassively multi-core architectures, namely Graphic Processor Units (GPUs). Theperformance of the proposed method in terms of accuracy of the reconstructionand computing time shows that the RBF interpolant can be very effective forsuch problem.
展开▼