Constructive generation of the medial axis for solid models

•A novel constructive method is given based on reuse of the existing MA.•Only the MA in the local re-dilation region needs to be regenerated.•Some new properties of MA are analyzed for the constructive MA generation.•The local re-dilation region is incrementally determined for Boolean subtraction operations.

The medial axis (MA) is a simplified representation of complicated models which is widely used in current research. However, the efficient generation of the MA for complicated solid models continues to pose a challenge. In this study, a constructive approach for the generation of the MA is proposed for solid models after they are voxelized. With this method, the MA of the model constructed from two operand models via a Boolean operation is efficiently generated by merging the MAs of the operand models in a certain way, instead of regenerating them from scratch. To support the proposed method, the affected region of the resultant model is computed first using a Boolean operation. Second, only the MA in the affected region is regenerated by distance dilation. Third, the complete MA of the resultant model is constructed by combining the newly generated MA with the unchanged MAs of the operand models. In this study, the accuracy and complexity are analyzed for the final MA and some examples are given to illustrate the performance of the proposed method.