Generation of hierarchical multi-resolution medial axis for CAD models

•An efficient approach to generating hierarchical multi-resolution MA of a CAD model is proposed.•The affected region and the re-voxelization region are proposed to reuse voxels when the MA level is upgraded.•An adaptive double queue distance dilation based algorithm is proposed to refine MA automatically.•The MA quality metric is proposed to evaluate the MA. Medial axis (MA) is used as an effective description for objects in many engineering fields. A difficulty for the current methods for the generation of MA of CAD models is the balance between the efficiency and the quality. In this study, an approach to iteratively generating hierarchical multi-resolution MA is proposed. In each iteration, only a small part of MA that affects MA quality is refined, by which the time cost and the space cost are reduced greatly. First, the model is voxelized and its initial MA is generated by distance dilation method. Meanwhile, the MA quality is computed and evaluated. Second, if the MA quality does not satisfy the requirement, upgrade the MA level and re-compute the local MA in the affected region until the MA quality does. Finally, by combining the local MA in the affected region with the reused MA in other regions, hierarchical multi-resolution MA is obtained. Several examples are given to demonstrate the outperformance of the proposed method in terms of time and space.