Parallel Generation of Medial Axis for Solid Models

Efficient generation of medial axis (MA) is still a great challenge for complicated solid models. In this study, a novel parallel method is proposed to generate MA of solid models with freeform surfaces efficiently after it is voxelized. Firstly, the input solid model is voxelized and divided into several blocks with nearly the same size. Then, the local MA voxels are calculated for each block based distance dilation method. Thirdly, the influence between different blocks is considered deliberately with the boundary information (BI) of other blocks to enable the local MA voxels to become a correct MA part of the whole model by complete inter-dilation (CID). Finally, Those MA voxels are combined and refined to become the resultant MA of the input model. The last three main processes can all be conducted efficiently in parallel. Several groups of experiments are given to illustrate the outperformance of the proposed method.